Data Migration To IBM Storage Systems PDF

Front cover
Data Migration to IBM

Disk Storage Systems
Highlights tool and techniques for Open
Systems and z/OS
Addresses appliance, host, and

storage-based migrations
Includes z/OS TDMF, TDMF

TCP/IP, and zDMF
Chris Seiwert
Peter Klee
Lisa Martinez
Max Pei
Mladen Portak
Alex Safonov
Edgar Strubel
Gabor Szabo
Ron Verbeek
ibm.com/redbooks
International Technical Support Organization
Data Migration to IBM Disk Storage Systems
February 2012
SG24-7432-01
Note: Before using this information and the product it supports, read the information in “Notices” on
page xi.
Second Edition (February 2012)
This edition applies to data migration products and techniques as of July 2011.
© Copyright International Business Machines Corporation 2011, 2012. All rights reserved.
Note to U.S. Government Users Restricted Rights -- Use, duplication or disclosure restricted by GSA ADP Schedule
Contract with IBM Corp.
Contents
Notices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi
Trademarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xii
Summary of changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii

February 2012, Second Edition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv
The team who wrote this book . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv
Now you can become a published author, too! . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii
Comments welcome. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xviii
Stay connected to IBM Redbooks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xviii
Part 1. Data migration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Chapter 1. Introducing disk data migration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

1.1 Data migration and its business reasons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2 Aspects of data migration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Addressing data migration issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Chapter 2. Migration techniques and processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

2.1 Data migration techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.1.1 Host-Based migration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.1.2 Array-Based migrations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.1.3 Appliance-Based migrations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.1.4 Migrating using backup and restore . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.1.5 Data migration in System z environments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.1.6 Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2 Key decision factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2.1 Migration performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2.2 Migration backout scenarios, protecting source and target volumes . . . . . . . . . . 14
2.2.3 Understand target storage requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2.4 Different source and target hardware . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.2.5 Application downtime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.2.6 Key decision factors summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.3 Data migration process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.3.1 Planning phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.3.2 Validation phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.4 Preparing DS8000 for data migration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.4.1 Available hardware resources. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.4.2 Understanding array characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
Chapter 3. IBM service offerings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

3.1 IBM Global Services offerings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.1.1 IBM Migration Services for data for Open Systems . . . . . . . . . . . . . . . . . . . . . . . 32
3.1.2 IBM Migration Services for System z data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.1.3 IBM Migration Services for Network Attached Storage Systems . . . . . . . . . . . . . 35
3.2 IBM Systems and Technology Group Lab Services . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.2.1 IBM XIV Migration Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.2.2 IBM DS8000 Data Migration Services. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
© Copyright IBM Corp. 2011, 2012. All rights reserved. iii

3.2.3 Additional Services & Offerings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
Part 2. Migration scenarios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
Chapter 4. Data migration using IBM Remote Mirror and Copy . . . . . . . . . . . . . . . . . . 41

4.1 IBM System Storage Remote Mirror and Copy overview . . . . . . . . . . . . . . . . . . . . . . . 42
4.1.1 Metro Mirror . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.1.2 Global Copy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.1.3 Target DS8000 configuration considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.1.4 Automation software: Tivoli Storage Productivity Center for Replication . . . . . . . 47
4.2 DS8000 user interface for Copy Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.2.1 IBM System Storage DS8000 Storage Manager . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.2.2 IBM System Storage DS8000 command-line interface. . . . . . . . . . . . . . . . . . . . . 49
4.2.3 System z interfaces. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.3 Data migration from Enterprise Storage Server Model 800 to DS8000 . . . . . . . . . . . . 51
4.3.1 Adding Enterprise Storage Server Copy Services Domain to the DS8000 Storage
Complex . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.3.2 Creating Remote Mirror and Copy paths . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.3.3 Creating Remote Mirror and Copy pairs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
4.3.4 Completing the data migration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
4.4 Data migration from DS8000 to DS8000. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
4.4.1 Configuring the remote DS8000 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
4.4.2 Creating Remote Mirror and Copy paths . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
4.4.3 Creating Remote Mirror and Copy pairs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
4.4.4 Completing the data migration from DS8000 to DS8000 . . . . . . . . . . . . . . . . . . . 85
4.5 Post-migration tasks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
4.5.1 AIX post-migration tasks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
4.5.2 Solaris and VxVM post-migration tasks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
4.5.3 System z post-migration tasks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
Chapter 5. DSCLIbroker . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105

5.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
5.1.1 DSCLIbroker concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
5.1.2 DSCLIbroker scripts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
5.1.3 User customized scripts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
5.1.4 Additional useful scripts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
5.2 System environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
5.2.1 Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
5.2.2 Communication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
5.2.3 Users. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
5.2.4 DSCLIbroker maintenance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
5.2.5 Licenses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
5.3 Programming techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
5.3.1 Using DSCLIbroker scripts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
5.3.2 Using DSCLIbroker libraries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
5.4 Automation techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
5.4.1 Subjects for writing automation scripts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
5.4.2 Generating DSCLI scripts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
5.4.3 Using control files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
5.5 Considerations about configuration data. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
5.5.1 Data sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
5.5.2 Importing data to the repository . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
5.6 Example migration using DSCLIbroker . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
5.6.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
iv Data Migration to IBM Disk Storage Systems

5.6.2 Analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
5.6.3 Test and develop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
5.6.4 Planning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
5.6.5 Running. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
5.6.6 Validating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
Chapter 6. IBM XIV data migration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133

6.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
6.2 Handling I/O requests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
6.2.1 Methods of handling write requests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
6.2.2 Multi-pathing with data migrations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
6.3 Data migration steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
6.3.1 Initial connection setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
6.3.2 Creating a data migration volume on XIV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
6.3.3 Activate a data migration on XIV. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
6.3.4 Defining the host on XIV and bringing the host online . . . . . . . . . . . . . . . . . . . . 149
6.3.5 Completing the data migration on XIV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
6.4 Command-line interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
6.4.1 Using XCLI scripts or batch files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
6.4.2 Sample scripts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
6.5 Manually creating the migration volume . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
6.6 Changing and monitoring the progress of a migration . . . . . . . . . . . . . . . . . . . . . . . . 158
6.6.1 Changing the synchronization rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
6.6.2 Monitoring migration speed. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
6.6.3 Monitoring migration using the XIV event log . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
6.6.4 Monitoring migration speed through the fabric . . . . . . . . . . . . . . . . . . . . . . . . . . 161
6.6.5 Monitoring migration speed through the non-XIV storage. . . . . . . . . . . . . . . . . . 161
6.7 Thick-to-thin migration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
6.7.1 Writing zeros to recover space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
6.7.2 Recovering space after the migration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
6.8 Resizing the XIV volume after migration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
6.9 Troubleshooting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
6.9.1 Target connectivity fails . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
6.9.2 Remote volume LUN is unavailable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
6.9.3 Local volume is not formatted . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
6.9.4 Host server cannot access the XIV migration volume. . . . . . . . . . . . . . . . . . . . . 168
6.9.5 Remote volume cannot be read . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
6.9.6 LUN is out of range . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
6.10 Backing out of a data migration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
6.10.1 Back out before migration is defined on the XIV . . . . . . . . . . . . . . . . . . . . . . . . 169
6.10.2 Back out after a data migration is defined but not activated . . . . . . . . . . . . . . . 169
6.10.3 Back out after a data migration is activated but is not complete . . . . . . . . . . . . 169
6.10.4 Back out after a data migration reaches the synchronized state . . . . . . . . . . . 170
6.11 Migration checklist. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
6.12 Other considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
Chapter 7. SAN Volume Controller-based migration . . . . . . . . . . . . . . . . . . . . . . . . . . 175

7.1 IBM System Storage SAN Volume Controller overview . . . . . . . . . . . . . . . . . . . . . . . 176
7.1.1 SAN Volume Controller components, concepts, and terminology . . . . . . . . . . . 177
7.1.2 SAN Volume Controller copy services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
7.1.3 Metro Mirror . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
7.1.4 Global Mirror . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
7.2 SAN Volume Controller concepts for migrating the data. . . . . . . . . . . . . . . . . . . . . . . 185
Contents v
7.2.1 Data migration using SAN Volume Controller volume migration. . . . . . . . . . . . . 185
7.2.2 Data migration using SAN Volume Controller FlashCopy. . . . . . . . . . . . . . . . . . 187
7.2.3 Data migration using SAN Volume Controller Metro Mirror . . . . . . . . . . . . . . . . 188
7.2.4 Data migration using mirrored volumes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
7.3 Migrating using SAN Volume Controller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
7.3.1 Migrating extents. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
7.3.2 Migrating extents off an MDisk that is being deleted. . . . . . . . . . . . . . . . . . . . . . 192
7.3.3 Migrating a volume between storage pools. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
7.3.4 Using volume mirroring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
7.3.5 Image mode volume migration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
7.3.6 Migrating the volume to image mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
7.3.7 Migrating a volume between I/O groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
7.3.8 Monitoring the migration progress. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
7.3.9 Parallelism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
7.3.10 Error handling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
7.3.11 Migration tips. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
7.4 SAN Volume Controller Migration preparation prerequisites. . . . . . . . . . . . . . . . . . . . 200
7.4.1 Fabric zoning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
7.4.2 Connect SAN Volume Controller to the fabric for migration . . . . . . . . . . . . . . . . 201
7.4.3 Remove SAN Volume Controller from the fabric after migration. . . . . . . . . . . . . 202
7.4.4 Back-End storage consideration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
7.4.5 Unsupported storage systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
7.4.6 Host attachment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
7.5 Migrating SAN disks to SAN Volume Controller volumes and back to SAN . . . . . . . . 203
7.5.1 Connecting the SAN Volume Controller to your SAN fabric . . . . . . . . . . . . . . . . 205
7.5.2 Preparing your SAN Volume Controller to virtualize disks . . . . . . . . . . . . . . . . . 206
7.5.3 Moving the LUNs to the SAN Volume Controller . . . . . . . . . . . . . . . . . . . . . . . . 210
7.5.4 Migrating image mode volumes to volumes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
7.5.5 Performance analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
7.5.6 Preparing to migrate from the SAN Volume Controller . . . . . . . . . . . . . . . . . . . . 220
7.5.7 Creating new LUNs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
7.5.8 Migrating the managed volumes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
7.5.9 Removing the LUNs from the SAN Volume Controller . . . . . . . . . . . . . . . . . . . . 225
7.6 SAN Volume Controller Volume migration between two storage pools . . . . . . . . . . . 227
7.6.1 Environment description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228
7.6.2 Performance measurement. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
7.6.3 Migration steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
7.6.4 Performance Analyses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
7.7 Data migration using SAN Volume Controller mirrored volumes . . . . . . . . . . . . . . . . 237
7.7.1 Environment description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
7.7.2 Creating mirrored volumes using the SAN Volume Controller GUI. . . . . . . . . . . 241
7.7.3 Creating mirrored volumes using CLI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
7.7.4 Performance analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246
7.8 Data migration using SAN Volume Controller Metro Mirror. . . . . . . . . . . . . . . . . . . . . 246
7.8.1 SAN Volume Controller Metro Mirror partnership . . . . . . . . . . . . . . . . . . . . . . . . 247
7.8.2 SAN Volume Controller Metro Mirror relationships . . . . . . . . . . . . . . . . . . . . . . . 249
7.8.3 Starting and monitoring SAN Volume Controller Metro Mirror Copy. . . . . . . . . . 254
7.8.4 Stopping SAN Volume Controller Metro Mirror Copy . . . . . . . . . . . . . . . . . . . . . 257
7.8.5 Performance overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
7.9 SAN Volume Controller as data migration engine. . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
7.10 Other resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
Chapter 8. Using mirroring techniques. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
vi Data Migration to IBM Disk Storage Systems

8.1 Concepts of the LVM. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266
8.2 Preparation and planning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
8.3 Migrating data using Windows Logical Disk Manager. . . . . . . . . . . . . . . . . . . . . . . . . 268
8.3.1 Sample Windows LDM migration test environment . . . . . . . . . . . . . . . . . . . . . . 268
8.3.2 High-level plan for migration using Windows LDM mirroring . . . . . . . . . . . . . . . 269
8.3.3 Detailed steps for migration using Windows LDM mirroring . . . . . . . . . . . . . . . . 271
8.4 Data migration using Solaris Volume Manager . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288
8.4.1 Test environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288
8.4.2 Solaris Volume Manager mirroring process . . . . . . . . . . . . . . . . . . . . . . . . . . . . 289
8.4.3 High-level plan for migration using SVM mirroring . . . . . . . . . . . . . . . . . . . . . . . 290
8.4.4 Detailed steps for migration using SVM mirroring. . . . . . . . . . . . . . . . . . . . . . . . 291
8.5 Data migration using Veritas Storage Foundation . . . . . . . . . . . . . . . . . . . . . . . . . . . 304
8.5.2 Veritas Volume Manager mirroring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
8.5.3 High-level plan for migration using VxVM mirroring . . . . . . . . . . . . . . . . . . . . . . 306
8.5.4 Detailed steps for migration using VxVM mirroring . . . . . . . . . . . . . . . . . . . . . . . 306
8.6 Data migration using HP-UX Volume Manager mirroring . . . . . . . . . . . . . . . . . . . . . . 320
8.6.1 High-level plan for migration using HP_UX Volume Manager mirroring . . . . . . . 321
8.6.2 Detailed steps for migration using HP-UX Volume Manager mirroring . . . . . . . . 321
8.7 Data migration using AIX LVM mirroring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 329
8.7.1 ESS to DS8000 migration under AIX LVM test environment . . . . . . . . . . . . . . . 330
8.7.2 High-level migration plan using the AIX LVM migratepv -l . . . . . . . . . . . . . . . . . 331
8.7.3 Detailed steps using migratepv -l . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333
8.7.4 High-level migration plan using AIX LVM mklvcopy and syncvg commands . . . 339
8.7.5 Detailed migration steps using mklvcopy and syncvg . . . . . . . . . . . . . . . . . . . 341
8.8 Data migration in a PowerHA clustered environment using AIX LVM mirroring . . . . . 349
8.8.1 PowerHA test environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 350
8.8.2 High-level migration plan using PowerHA C-SPOC LVM mirroring . . . . . . . . . . 354
8.8.3 Detailed migration steps using PowerHA C-SPOC LVM mirroring . . . . . . . . . . . 355
8.8.4 Migration Backout Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 375
8.9 Data migration using Linux LVM2 mirroring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 376
8.9.1 Introduction to Linux LVM2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 376
8.9.2 Multipathing considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 376
8.9.3 Linux LVM migration scenario. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 377
8.10 Data migration using network block devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 389
8.10.1 General architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 389
8.10.2 Using shell commands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391
8.10.3 Using software RAID1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391
8.10.4 Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393
8.10.5 Using Linux LVM2 mirroring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393
Chapter 9. Using TDMF for z/OS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 395

9.1 TDMF for z/OS overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 396
9.1.1 Data migration and migration tool definitions and characteristics . . . . . . . . . . . . 396
9.1.2 Terminology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 396
9.1.3 TDMF z/OS components architecture. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 397
9.1.4 TDMF z/OS process flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 398
9.1.5 TDMF z/OS hardware compatibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 401
9.1.6 Installing TDMF z/OS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 402
9.1.7 TDMF z/OS customer requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403
9.2 Customer scenarios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 405
9.2.2 Migration steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 408
Contents vii
9.3 TDMF preferred practices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 418
9.3.1 Keeping current. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 418
9.3.2 Setting default options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 418
9.3.3 Storage requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 419
9.3.4 Communications data set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 419
9.3.5 Participation of agent systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 420
9.3.6 Protection of target volume data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 421
9.3.7 Pacing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 421
9.3.8 Rank contention and storage subsystem performance. . . . . . . . . . . . . . . . . . . . 422
9.3.9 TDMF interaction with other programs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 422
9.3.10 Identification of volumes requiring special handling . . . . . . . . . . . . . . . . . . . . . 423
9.3.11 Migration considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 425
9.3.12 Estimating how long it takes to move the data . . . . . . . . . . . . . . . . . . . . . . . . . 428
9.3.13 Terminating a TDMF session . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 431
Chapter 10. z/OS Dataset Mobility Facility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 433

10.1 z/OS Dataset Mobility Facility overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434
10.2 zDMF migration process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434
10.2.1 zDMF migration steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 435
10.2.2 Planning the migration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 435
10.2.3 Pre migration tasks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 436
10.2.4 Migration phase using DFSMS, HSM and utilities, or both . . . . . . . . . . . . . . . . 437
10.2.5 zDMF migration phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 438
10.2.6 Typical migration process scenario. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 442
10.2.7 Migration performance and scalability. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 445
10.2.8 Post-migration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 449
10.3 Important zDMF-related topics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 450
10.3.1 Understanding the catalog structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 450
10.3.2 BCS record types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 451
10.3.3 Catalog diagnostic recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 452
10.3.4 Using Global Resource Serialization (GRS) . . . . . . . . . . . . . . . . . . . . . . . . . . . 454
10.3.5 Ring complex . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 454
10.3.6 Star complex . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 456
10.3.7 Reserve handling requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 456
10.4 Example migration in a shared DB2 test environment . . . . . . . . . . . . . . . . . . . . . . . 458
10.4.1 Test Environment Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 459
10.4.2 Preparing DSN List for Test Migration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 462
10.4.3 Creating BATCH JCL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 463
10.4.4 Checking Data and Starting DB2 Load . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 465
10.4.5 Starting and Monitoring Test Migration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 466
10.4.6 Diverting the Group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 472
10.4.7 Starting post migration cleanup procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . 478
Chapter 11. Using TDMF TCP/IP for z/OS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 483

11.1 TDMF TCP/IP for z/OS overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 484
11.1.1 Long-distance data migration software solution . . . . . . . . . . . . . . . . . . . . . . . . 484
11.1.2 Long-distance data migration characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . 484
11.1.3 Terminology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 485
11.1.4 TDMF z/OS components architecture. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 486
11.1.5 TDMF z/OS hardware compatibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 487
11.1.6 Install TDMF TCP/IP z/OS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 487
11.1.7 TDMF z/OS customer requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 488
11.2 TDMF general guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 488
viii Data Migration to IBM Disk Storage Systems

11.2.1 Keeping current. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 488
11.2.2 Creating a consistent data structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 489
11.2.3 Summary of replication tasks and steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 490
11.2.4 Link performance information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 494
11.3 TDMF TCP/IP example replication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 496
11.3.1 Overview of replication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 497
11.3.2 Naming conventions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 497
11.3.3 TDMF TCP/IP examples for a network test. . . . . . . . . . . . . . . . . . . . . . . . . . . . 507
11.3.4 TDMF TCP/IP performance example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 512
11.3.5 TCP/IP information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 512
Appendix A. Network block devices for Linux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 515

Network block devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 516
Suggested methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 516
Using software RAID1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 517
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 518
Appendix B. CLI Conversion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 521

Migrating ESS Copy Services tasks to DS8000 CLI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 522
Review the ESS tasks to migrate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 522
Convert the individual tasks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 526
Copy Services commands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 529
Copy Services notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 530
Related publications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 531

IBM Redbooks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 531
Other publications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 531
Online resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 531
Help from IBM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 531
...................................................................... 532
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 533
Contents ix
x Data Migration to IBM Disk Storage Systems
Notices
This information was developed for products and services offered in the U.S.A.
IBM may not offer the products, services, or features discussed in this document in other countries. Consult
your local IBM representative for information on the products and services currently available in your area. Any
reference to an IBM product, program, or service is not intended to state or imply that only that IBM product,
program, or service may be used. Any functionally equivalent product, program, or service that does not
infringe any IBM intellectual property right may be used instead. However, it is the user's responsibility to
evaluate and verify the operation of any non-IBM product, program, or service.
IBM may have patents or pending patent applications covering subject matter described in this document. The
furnishing of this document does not give you any license to these patents. You can send license inquiries, in
writing, to:
IBM Director of Licensing, IBM Corporation, North Castle Drive, Armonk, NY 10504-1785 U.S.A.
The following paragraph does not apply to the United Kingdom or any other country where such
provisions are inconsistent with local law: INTERNATIONAL BUSINESS MACHINES CORPORATION
PROVIDES THIS PUBLICATION "AS IS" WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESS OR
IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF NON-INFRINGEMENT,
MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Some states do not allow disclaimer of
express or implied warranties in certain transactions, therefore, this statement may not apply to you.
This information could include technical inaccuracies or typographical errors. Changes are periodically made
to the information herein; these changes will be incorporated in new editions of the publication. IBM may make
improvements and/or changes in the product(s) and/or the program(s) described in this publication at any time
without notice.
Any references in this information to non-IBM websites are provided for convenience only and do not in any
manner serve as an endorsement of those websites. The materials at those websites are not part of the
materials for this IBM product and use of those websites is at your own risk.
IBM may use or distribute any of the information you supply in any way it believes appropriate without incurring
any obligation to you.
Information concerning non-IBM products was obtained from the suppliers of those products, their published
announcements or other publicly available sources. IBM has not tested those products and cannot confirm the
accuracy of performance, compatibility or any other claims related to non-IBM products. Questions on the
capabilities of non-IBM products should be addressed to the suppliers of those products.
This information contains examples of data and reports used in daily business operations. To illustrate them
as completely as possible, the examples include the names of individuals, companies, brands, and products.
All of these names are fictitious and any similarity to the names and addresses used by an actual business
enterprise is entirely coincidental.
COPYRIGHT LICENSE:
This information contains sample application programs in source language, which illustrate programming
techniques on various operating platforms. You may copy, modify, and distribute these sample programs in
any form without payment to IBM, for the purposes of developing, using, marketing or distributing application
programs conforming to the application programming interface for the operating platform for which the sample
programs are written. These examples have not been thoroughly tested under all conditions. IBM, therefore,
cannot guarantee or imply reliability, serviceability, or function of these programs.
© Copyright IBM Corp. 2011, 2012. All rights reserved. xi

Trademarks
IBM, the IBM logo, and ibm.com are trademarks or registered trademarks of International Business Machines
Corporation in the United States, other countries, or both. These and other IBM trademarked terms are
marked on their first occurrence in this information with the appropriate symbol (® or ™), indicating US
registered or common law trademarks owned by IBM at the time this information was published. Such
trademarks may also be registered or common law trademarks in other countries. A current list of IBM
trademarks is available on the Web at http://www.ibm.com/legal/copytrade.shtml
The following terms are trademarks of the International Business Machines Corporation in the United States,
other countries, or both:
AIX® IBM® RMF™
BladeCenter® IMS™ S/390®
CICS® Informix® Storwize®
DB2® Lotus Notes® System i®
DS4000® Lotus® System Storage DS®
DS6000™ MVS™ System Storage®
DS8000® Notes® System z®
Easy Tier® OS/390® TDMF®
Enterprise Storage Server® Parallel Sysplex® Tivoli®
ESCON® PowerHA® VM/ESA®
eServer™ PR/SM™ XIV®
FICON® pSeries® z/OS®
FlashCopy® RACF® z/VM®
GDPS® Rational® z/VSE®
Global Technology Services® Redbooks® zSeries®
HACMP™ Redbooks (logo) ®
The following terms are trademarks of other companies:
Microsoft, Windows, and the Windows logo are trademarks of Microsoft Corporation in the United States,
other countries, or both.
Snapshot, and the NetApp logo are trademarks or registered trademarks of NetApp, Inc. in the U.S. and other
countries.
UNIX is a registered trademark of The Open Group in the United States and other countries.
Disk Magic, and the IntelliMagic logo are trademarks of IntelliMagic BV in the United States, other countries,
or both.
QLogic, and the QLogic logo are registered trademarks of QLogic Corporation. SANblade is a registered
trademark in the United States.
Intel, Intel logo, Intel Inside, Intel Inside logo, Intel Centrino, Intel Centrino logo, Celeron, Intel Xeon, Intel
SpeedStep, Itanium, and Pentium are trademarks or registered trademarks of Intel Corporation or its
subsidiaries in the United States and other countries.
Linux is a trademark of Linus Torvalds in the United States, other countries, or both.
Other company, product, or service names may be trademarks or service marks of others.
xii Data Migration to IBM Disk Storage Systems

Summary of changes
This section describes the technical changes made in this edition of the book. This edition
might also include minor corrections and editorial changes that are not identified.
Summary of Changes
for SG24-7432-01
for Data Migration to IBM Disk Storage Systems
as created or updated on February 23, 2012.
February 2012, Second Edition

This revision reflects the addition, deletion, or modification of new and changed information
described below.
New information
򐂰 Data migration in IBM PowerHA® clustered environments
򐂰 IBM XIV® data migration
򐂰 IBM® DSCLIbroker scripting framework
򐂰 IBM TDMF® TCP/IP for IBM z/OS®
Changed information
򐂰 Screen captures updated to reflect latest software available at the time of writing
© Copyright IBM Corp. 2011, 2012. All rights reserved. xiii

xiv Data Migration to IBM Disk Storage Systems
Preface
Data migration has become a mandatory and regular activity for most data centers.
Companies need to migrate data not only when technology needs to be replaced, but also for
consolidation, load balancing, and disaster recovery.
This IBM Redbooks® publication addresses the aspects of data migration efforts while
focusing on the IBM System Storage® as the target system. Data migration is a critical and
complex operation, and this book provides the phases and steps to ensure a smooth
migration. Topics range from planning and preparation to execution and validation.
The book also reviews products and describes available IBM data migration services
offerings. It explains, from a generic standpoint, the appliance-based, storage-based, and
host-based techniques that can be used to accomplish the migration. Each method is
explained including the use of the various products and techniques with different migration
scenarios and various operating system platforms.
This document targets storage administrators, storage network administrators, system

designers, architects, and IT professionals who design, administer or plan data migrations in
large data Centers. The aim is to ensure that you are aware of the current thinking, methods,
tools, and products that IBM can make available to you. These items are provided to ensure a
data migration process that is as efficient and problem-free as possible.
The material presented in this book was developed with versions of the referenced products
as of June 2011.
The team who wrote this book

This IBM Redbooks publication was produced by a team of specialists from around the world
working at the International Technical Support Organization, San Jose Center.
Chris Seiwert is an IT Specialist in the IBM European Storage Competence Center (ESCC),
Mainz, Germany. He has 12 years of experience in SAN, High End Storage, Data Center
Analysis and Planning, and 19 years in computer science. He holds a Bachelor of Science
degree in Computer Science, and his areas of expertise also include Data Migration, JavaTM
development, and HA Cluster solutions. Chris filed his first patent in 2005 for an IBM Data
Migration tool. He has written about SAN and data migration in IBM internal and external
publications, and has delivered presentations and workshops about these topics. He has
co-authored three previous IBM Redbooks and was the ITSO lead for this book.
Peter Klee is an IBM Professional Certified IT specialist in IBM Germany. He has 17 years of
experience in Open Systems platforms, SAN networks, and high end storage systems in
huge data centers. He formerly worked in a large bank in Germany where he was responsible
for the architecture and the implementation of the disk storage environment. This environment
has included the installation from various storage vendors, including EMC, HDS, and IBM. He
joined IBM in 2003, where he worked for Strategic Outsourcing. Since July 2004, he has
worked for ATS System Storage Europe in Mainz. His main focus is Copy Services, Disaster
Recovery, and storage architectures for IBM DS8000® in the open systems environment.
Lisa Martinez is a storage architect in the Speciality Services Area in GTS. She has been in
this position since the beginning of 2011. In 2010 Lisa took a temporary assignment as a
Global Support Manager for Cardinal Health. Prior to this assignment she was a test architect
© Copyright IBM Corp. 2011, 2012. All rights reserved. xv

in disk storage focusing on system level test for XiV for three years, and copy services for
DS8K. Lisa holds degrees in Computer Science from New Mexico Highlands University and
Electrical Engineering from the University of New Mexico. She has been employed with IBM
for 14 years.
Max Pei is an Infrastructure Architect for GTS in IBM Canada, specializing in SAN, Storage,
and Backup systems. He has 14 years of experience in the IT industry and has been with IBM
since 2008. He holds a degree in Metallurgical Engineering. His areas of expertise include
planning and implementation of midrange and enterprise storage, storage networks, backup
systems, data migration and server virtualization.
Mladen Portak is a Client Technical Storage Specialist for STG in IBM Croatia, and
specializes in Storage systems. He is certified on the IBM Midrange and Enterprise Storage
Systems and Microsoft MCP. Mladen has 16 years of experience in the IT industry and has
been with IBM since 2008. Before joining IBM he worked on customer side as a team leader
responsible for virtualization solutions. His current area of expertise includes the planning,
architecture and implementation of midrange and enterprise storage, and storage area
networks for Open Systems.
Alex Safonov is a Senior IT Specialist with System Sales Implementation Services, IBM
Global Technology Services® Canada. He has over 20 years of experience in the computing
industry, with the last 15 years spent working on Storage and UNIX solutions. He holds
multiple product and industry certifications, including IBM Tivoli® Storage Manager, IBM
AIX®, and SNIA. Alexander spends most of his client contracting time working with Tivoli
Storage Manager, data archiving, storage virtualization, and replication and migration of data.
He holds an M.S. Honors degree in Mechanical Engineering from the National Aviation
University of Ukraine.
Edgar Strubel is a Server and Storage Specialist with the STG LAB Services Europe, Mainz,
Germany. He started working at IBM at the end of 2000 (Mainz Briefing Center, ATS Team)
and since 2002 he has been involved in online data migration in IBM zSeries® environments.
Prior to joining IBM, starting in 1980, Edgar worked at BASF/COMPAREX in mainframe IT
environments, performing service and support in hardware and software for printers, tapes,
libraries, disks, and processors.
Gabor Szabo is an IBM Certified Solution Designer working as Test Engineer team leader of
DS8000 Development Support team in IBM Hungary, Vac. He has 8 years of experience in
the IT industry and has been with IBM since 2005. Gabor current area of expertise includes
the high-end storage system testing, test optimization, and new product implementation.
Ron Verbeek is a Senior Consulting IT Specialist with Data Center Services, IBM Global
Technology Services Canada. He has over 23 years of experience in the computing industry
with IBM, with the last 11 years working on Storage and Data solution services. He holds
multiple product and industry certifications, including SNIA Storage Architect, and has
co-authored one previous IBM Redbooks on the IBM XIV Storage Subsystem. Ron spends
most of his client time in technical pre-sales solutions, defining and creating storage and
optimization solutions. He has extensive experience in data transformation services and
information life cycle consulting. He holds a Bachelor of Science degree in Mathematics from
McMaster University in Canada.
xvi Data Migration to IBM Disk Storage Systems

Figure 1 The team who wrote this book
From left to right: Bert Dufrasne, Edgar Strubel, Chris Seiwert, Alex Savonov, Ron Verbeek,
Peter Klee, Mladen Portak, Gabor Szabo, Max Pei. Missing: Lisa Martinez.
Special thanks to Uwe Heinrich Mueller, Uwe Schweikhard, and Mike Schneider
IBM Lab services Mainz, for excellent lab support during the residency.
Thanks to the following people for their contributions to this project:
Bernd Mueller, Michael Murtagh, Ralf Wohlfarth

IBM Systems & Technology Group Mainz, Europe
Lori Bideaux and Bert Dufrasne

IBM International Technical Support Organization (ITSO)
Michael Moss
Technical Support Softek (an IBM Company)
Now you can become a published author, too!

Here’s an opportunity to spotlight your skills, grow your career, and become a published
author—all at the same time! Join an ITSO residency project and help write a book in your
area of expertise, while honing your experience using leading-edge technologies. Your efforts
will help to increase product acceptance and customer satisfaction, as you expand your
network of technical contacts and relationships. Residencies run from two to six weeks in
length, and you can participate either in person or as a remote resident working from your
home base.
Find out more about the residency program, browse the residency index, and apply online at:
ibm.com/redbooks/residencies.html
Preface xvii
Comments welcome
Your comments are important to us!
We want our books to be as helpful as possible. Send us your comments about this book or
other IBM Redbooks publications in one of the following ways:
򐂰 Use the online Contact us review Redbooks form found at:
ibm.com/redbooks
򐂰 Send your comments in an email to:
[email protected]
򐂰 Mail your comments to:
IBM Corporation, International Technical Support Organization
Dept. HYTD Mail Station P099
2455 South Road
Poughkeepsie, NY 12601-5400
Stay connected to IBM Redbooks

򐂰 Find us on Facebook:
http://www.facebook.com/IBMRedbooks
򐂰 Follow us on Twitter:
http://twitter.com/ibmredbooks
򐂰 Look for us on LinkedIn:
http://www.linkedin.com/groups?home=&gid=2130806
򐂰 Explore new Redbooks publications, residencies, and workshops with the IBM Redbooks
weekly newsletter:
https://www.redbooks.ibm.com/Redbooks.nsf/subscribe?OpenForm
򐂰 Stay current on recent Redbooks publications with RSS Feeds:
http://www.redbooks.ibm.com/rss.html
xviii Data Migration to IBM Disk Storage Systems

Part 1
Part 1 Data migration

This part defines data migration and reviews the various business reasons for migrating data.
It explores the advantages and disadvantages of the various migration techniques, and
highlights some of the available migration products. It also includes the phases of a migration
project and available IBM disk migration services.
This part contains the following chapters:

򐂰 Introducing disk data migration
򐂰 Migration techniques and processes
򐂰 IBM service offerings
© Copyright IBM Corp. 2011, 2012. All rights reserved. 1

2 Data Migration to IBM Disk Storage Systems
1
Chapter 1. Introducing disk data migration

This chapter defines disk data migration and examines the business reasons for migrating
data. It contains the following sections:
򐂰 Data migration and its business reasons
򐂰 Aspects of data migration
򐂰 Addressing data migration issues

1.1 Data migration and its business reasons
Data migration is the transferring of data between storage subsystem types, formats, or
computer systems. These systems can be single or multiple, and similar or dissimilar.
Migration is usually performed under program control to try to make movement of data as
complete and as automated as is possible.
Data migration is needed when you change computer systems or upgrade to new products.
As with any data center change, you want to avoid disrupting or disabling active applications.
You might migrate data for the following reasons:

򐂰 Your organization wants to use new storage devices to increase the size of your online
storage or for more flexibility.
򐂰 You want to resolve performance issues.
򐂰 You need to physically reduce the footprint of your storage subsystem within the data
center and release space. Reducing the footprint reduces the costs of power
consumption, air conditioning, and lighting.
򐂰 You want to employ data migration technologies to implement a disaster recovery solution.
򐂰 You need the new functions and facilities offered by evolving technology to stay
competitive.
򐂰 You need to relocate your data center.
1.2 Aspects of data migration

Migrating data has the following general concerns, which require that you take a holistic view
of your data center:
򐂰 Configuration: New hardware might need to be configured into the systems being migrated
before you can start.
򐂰 Operations: A time needs to be found when the migration can be performed. Scheduling is
especially important if you use offline data migration techniques that require the
applications to be stopped. After the migration, your data center operating procedures will
need to be reviewed and updated.
򐂰 Data: Not all data can be migrated in the same way. For example, local paging data sets
on IBM OS/390® cannot be moved directly. The restriction is because manual handling of
local paging data sets is required to keep OS/390 systems up and running.
򐂰 Infrastructure: Changes to infrastructure might be required to accommodate new storage
subsystem technologies. For example, new technology can require new attachment
methods or enhanced data rate capabilities. The most common upgrades are from Escon
to Ficon attachment, and for Open Systems installing a SAN Fabric or upgrading your
existing fabric.
򐂰 Applications: You might need to modify or add installation parameters in your applications
to accommodate new storage subsystem devices or infrastructures.
򐂰 Performance: Migration of data is an opportunity to maximize performance by carefully
relocating high use data. However, if you are not careful about where you migrate data,
you can create performance issues.

1.3 Addressing data migration issues
All of the above scenarios are fairly commonplace in modern IT operations and in virtually all
businesses. However, even these routine practices can cause problems for IT managers,
database administrators, and their teams. The problems tend to include the following issues:
򐂰 Extended or unexpected downtime
򐂰 Data corruption, missing data, and data loss
򐂰 Application performance issues
򐂰 Technical compatibility issues
To avoid impact to business operations, most data migration projects are scheduled during
off-hours, primarily during weekends and over extended holiday periods. However, this delay
can increase migration costs because of staff overtime and negatively impact staff morale.
Furthermore, taking systems down for migration, even over a weekend period, can severely
affect business operations. The impact is especially severe if you cannot bring the systems
back up in time for the online day.
These potential problems cause some organizations to significantly delay purchasing new
technology, or even to delay the deployment of already purchased technology. These delays
cause further problems because older hardware can require more hands-on maintenance,
generally has lower performance, and is inherently more prone to failure.
You buy and deploy new technology to eliminate these issues. Therefore delays in
implementing new technology increases business risk because of your need to run
around-the-clock 24x7 applications with ever-shrinking batch windows. In addition, delaying
deployment of an already purchased or leased storage device raises its effective cost
because you are paying for both old and new devices.
Data migration can be low risk. Current IBM data migration technologies allow you to perform
most migrations with no downtime. IPL or server restarts are not always required, and no
volumes need to be taken offline. However, you might have to perform a scheduled IPL or
restart of a server if you are adding new equipment or applying system maintenance. In
addition, the latest migration software tools allow nondisruptive migration, allowing
applications to remain online during data movement without significant performance delays.
Chapter 1. Introducing disk data migration 5

2
Chapter 2. Migration techniques and

processes
This chapter reviews the data migration techniques and highlights how to select the
appropriate technique.
The chapter also includes a summary of the migration process based on a three-phase
approach:
򐂰 Planning phase
򐂰 Migration phase
򐂰 Validation phase
This chapter includes the following sections:

򐂰 Data migration techniques
򐂰 Key decision factors
򐂰 Data migration process
򐂰 Preparing DS8000 for data migration

2.1 Data migration techniques
Migrating data is always a disruptive process. Every migration technique affects the normal
operations of the system.
Selecting the appropriate technique depends on the criticality of the data being moved, the
resources available, and other business constraints and requirements. The different
techniques have different risks. Select the technique that provides the best combination of
efficiency of migration and low impact to system and users.
Various software products and tools can be used to migrate data:

򐂰 Volume management products
򐂰 Host-Based replication products
򐂰 Array-Based replication products
򐂰 Relocation utilities
򐂰 Custom-Developed scripts
Each product or tool has strengths and weaknesses. These include impact on performance,
operating system support, storage vendor platform support, and whether application
downtime is required to migrate data. Select the best product or tool for your needs.
Determine whether you can migrate data online or offline:

򐂰 Online migrations: Data can be copied from one set of volumes to another with no impact
to users and applications. In most cases, some preparation needs to be done on a system
before data can be copied nondisruptively. Examples include an upgrade of HBA device
drivers and replacement of multipathing software. These preparation steps often require
downtime because a reboot is needed to activate the installed code. After the preparations
are completed, the data can be moved or copied nondisruptively with little or no
performance impact.
򐂰 Offline migrations: The data must be in a known consistent state before they can be
copied or moved. The data must also remain unchanged for the duration of the migration
process. Depending on the amount you are migrating, data can remain unavailable for an
extended time, from several hours to days.
2.1.1 Host-Based migration

Host-Based migrations use functions provided by the following tools:
򐂰 Host operating system (OS) such as Logical Volume Manager (LVM for UNIX and Linux,
LDM for Windows)
򐂰 Add-on software such as IBM TDMF or Veritas Volume Manager (VxVM)
򐂰 Other native volume management tools.
Volume management software

Volume management software provides a simple and robust set of tools for selective copying,
mirroring, and migration of data. When you perform data migration on critical production
systems, you can use volume management software to tune data copying rate to minimize the
performance impact.
AIX, Solaris, HP-UX, Linux, Windows, and IBM z/OS are equipped with volume management
software that manages disk storage devices. You can use this software to configure and
manage volumes, file systems, paging, and swap spaces.

For a detailed description about how to perform these types of migrations, see Chapter 8,
“Using mirroring techniques” on page 265.
File copy
If the data you are migrating is a group of individual files and no volume management
software is available, use a file-level migration technique. This technique uses native OS or
third-party utilities and commands that support the file copy feature. Using copy commands
and utilities is the simplest form of data migration between two file systems.
In UNIX, you can use the following commands:

򐂰 cp
򐂰 cpio
򐂰 tar
򐂰 pax
򐂰 dump
򐂰 backup and restore
򐂰 rsync
In Windows, you can use the following commands:

򐂰 cp
򐂰 scopy
򐂰 xcopy
򐂰 robocopy
򐂰 Windows Explorer drag and drop function
File copy has the following advantages:

򐂰 Is available on every OS type that can be attached to IBM disk subsystems
򐂰 Can be used to copy between file systems with different sizes, so it can be used for file
system and volume consolidation
File copy has the following disadvantages:

򐂰 Works only with data organized in file systems format
򐂰 To preserve data consistency, all writes to a file must be stopped for the duration of the
copy process
򐂰 Some copy commands cannot preserve advanced metadata, such as access control lists
or permissions
Important: If your storage systems are attached through multiple paths, verify that the
multipath drivers for the old and new storage can coexist. If they cannot, revert the host to
a single path configuration and remove the incompatible driver before attaching the new
storage system.
Raw devices copy

Data on raw volumes can be read and written directly from and to a volume using tools such
as dd (a UNIX utility). Such tools are typically used for copying contents of a raw volume from
one physical device to another. Take care when an individual volume has external
dependencies. An example is a journal file system on one volume that has a corresponding
journal on another volume. In this case, you must copy the content of both volumes at the
same time. In addition, the configuration of the file systems must be adjusted to reflect the
name of the new volumes.
Chapter 2. Migration techniques and processes 9

Raw device copy commands such as dd provide a powerful interface for copying data
between volumes. However, you must ensure that target volume has sufficient space. You
must also update all configuration information to reflect the new volume name as in the
example.
The raw device copying method is an offline migration method. Applications accessing data
on raw logical volumes must remain offline for the duration of the data copying process. The
tools do not prevent you from reading data from a volume being used by an application. This
might result in inconsistent data on the target volume.
2.1.2 Array-Based migrations

An array-based migration uses IBM Copy Services functions to transfer data from one storage
subsystem to another. You can use functions provided by Metro Mirror, Global Mirror, and
Global Copy to replicate data between compatible storage arrays. This method is addressed
in more detail in Chapter 4, “Data migration using IBM Remote Mirror and Copy” on page 41,
and DS8000 Copy Services for IBM System z, SG24-6787.
2.1.3 Appliance-Based migrations

Appliance-Based data migrations use a virtualization device between a server and a disk
storage subsystem. This virtualization device acts as a mediator for block level or file level
protocols. The virtualization device allows you to migrate data from one physical device to
another without any impact on applications and users.
The appliance-based migrations addressed in this book are based on the SAN Volume
Controller for block level and f5 ARX-VE for file level storage.
SAN Volume Controller-Based migrations

IBM SAN Volume Controller can be used to migrate data between any supported back-end
disk storage platforms. To connect to SAN Volume Controller-Managed storage, install and
configure the compatible device driver on a server. After a server is connected to SAN
Volume Controller managed storage, all data migration activities are done by SAN Volume
Controller. There is no interaction with the server software.
Important: SAN Volume Controller-Based migrations can be used only for fixed block type
storage type of open systems. CKD storage used by IBM System z® is not supported by
SAN Volume Controller.
2.1.4 Migrating using backup and restore

Standard UNIX commands can be used to migrate data. The following variations can be
used:
򐂰 Back up and restore
򐂰 Dump and restore
򐂰 Other commands
The backup and restore options allow for consolidation because the tools are aware of the
data structures they handle.
These methods are unusual in that they do not require the source and target storage systems
to be connected to the host at the same time.

Back up and restore
In this variation, you back data up to an external device, such as tape, and restore it to the
target location. This method is slow because an extra copy step is introduced and every file or
block of data is transferred twice:
򐂰 From source to a temporary location
򐂰 From the temporary location to the target.
However, if you must remove the old storage system before installing the new one, you must
use an external storage device.
Migrating data using backup and restore generally have the most impact on system usage.
This process requires that applications and in certain cases file systems be in quiescent
states to ensure data consistency.
Dump and restore

These commands are similar to the Backup and Restore command. You can find them on
almost all forms of UNIX. This method in most cases also requires an intermediate device.
Other commands
You can find other commands on UNIX systems for backing up data. Again, these commands
require that you create an intermediate backup image of an object before restoring to a target
location.
You might not be able to use the volume management methods of migrating data in the
following cases:
򐂰 For databases that use their own storage administration software for managing raw
volumes
򐂰 Specialized applications that use raw volumes
In some cases, applications use volume names or serial numbers to generate license keys,
effectively becoming location dependent. These applications might not tolerate data migration
done with tools other than the tools supplied with the application itself such as data
export/import tools.
All open systems platforms and many applications provide native backup and restore
capabilities. They might not be sophisticated, but they are often suitable in smaller
environments. In large data centers, it is customary to have a common backup solution
across all systems. Either solution can be used for data migration.
The most common software packages that provide this function are:
򐂰 IBM Tivoli Storage Manager
򐂰 Legato Networker
򐂰 BrightStor ARCserve
򐂰 Symantec NetBackup
2.1.5 Data migration in System z environments

This section highlights methods for migrating data from existing disk storage systems onto the
DS8000, specifically in a System z environment. Some are similar to the techniques
previously explained. The following topics are addressed:
򐂰 Data migration based on physical volume migration
򐂰 Data migration based on logical data set migration

򐂰 Combination of physical and logical data migration
򐂰 Copy Services-based migration
Data migration based on physical volume migration

This method uses physical full volume migration, which requires the same device geometry
on the source and target volumes. The device geometry is defined by the track capacity and
the number of tracks per cylinder. Usually this is not an issue because over time the device
geometry of the IBM 3390 volume has become standardized. For organizations still using
other device geometry (for example, 3380), consider a device geometry conversion, if
possible. Conversion requires moving the data on a logical level (data set level), which allows
a reblocking during the migration from 3380 to 3390.
Physical full volume migration is possible with the following tools:

򐂰 Software-based:
– DFSMSdss
– TDMF
– FDRPAS
– z/OS Global Mirror (XRC)
Note: XRC requires a license on the DS8000.
򐂰 Storage-based:
– IBM Rational® Method Composer functions
Data migration based on logical data set migration

Data logical migration is a migration by data set that maintains catalog entries according to
the data movement between volumes. It is not a volume-based migration. Data logical
migration is the cleanest way to migrate data, and allows device conversion from, for example,
3380 to 3390. It also transparently supports multivolume data sets. Logical data migration is a
software-only approach, and does not rely on volume characteristics or on-device
geometries.
The following software products and components can be used for data logical migrations:
򐂰 DFSMS allocation management.
򐂰 Softek zDMF.
򐂰 System utilities such as:
– IDCAMS using the REPRO and EXPORT/IMPORT commands.
– IEBCOPY for Partitioned Data Sets (PDS) or Partitioned Data Sets Extended (PDSE).
– ICEGENER as part of DFSORT, which can handle sequential data but not VSAM data
sets.
– IEBGENER, which has the same restrictions as ICEGENER.
򐂰 Database utilities for data that are managed by certain database managers, such as IBM
DB2® or IBM IMS™. IBM CICS® as a transaction manager usually uses VSAM data sets.
Combination of physical and logical data migration

The following approach combines physical and logical data migration:
1. Perform a physical full volume copy to a larger capacity volume when both volumes have
the same device geometry. They must have the same track size and the same number of
tracks per cylinder.

2. Use COPYVOLID to keep the original volume label, and maintain catalog management.
You can still locate the data on the target volume through a standard catalog search.
3. Adjust the VTOC of the target volume to make the larger volume size visible to the system.
Use the ICKDSF REFORMAT command REFVTPC to refresh the VTOC, or use the
command EXTVTOC to expand the VTOC.
Tip: Issuing the EXTVTOC command requires you to delete and rebuild the VTOC
index using EXTINDEX in the REFORMAT command
4. Perform the logical data set copy operation to the larger volumes. This operation allows
you to use either DFSMSdss logical copy operations or the system-managed data
approach.
When there are no more data moves because the remaining data sets are in continual use,
schedule downtime to move them. You might have to run DFSMSdss jobs from a system that
has no active allocations on the volumes that need to be emptied.
2.1.6 Summary
Each method of data migration has strengths and limitations. Table 2-1 lists the pros and cons
of the suggested products and techniques covered in this book. It is not, however, a
comprehensive list of every technology and technique.
Table 2-1 An overview of migration techniques
Migration technique Pros Cons
Host-based 򐂰 Generally lowest initial 򐂰 Consumes host resources

򐂰 LVM implementation costs 򐂰 Operating system specific
򐂰 LDM 򐂰 Uses existing hosts and IP 򐂰 Management can become
򐂰 Add-on software such as network complex and time
VxVM 򐂰 LVM and LDM are parts of consuming
򐂰 Volume (block) level (Softek the operating system 򐂰 No central management
TDMF) or file (dataset) level 򐂰 Storage-vendor-agnostic console
(zDMF) 򐂰 Uses existing operating 򐂰 Might cause an initial
system skills outage to install the utility or
򐂰 Migration can happen software
online during peak hours
Network-based replication 򐂰 Supports heterogeneous 򐂰 Higher initial cost due to

򐂰 Softek TDMF (IP) environments (servers and hardware and replication
storage) software
򐂰 Single point of 򐂰 Requires proprietary
management for replication hardware
services 򐂰 Might require
򐂰 Does not consume host implementation of storage
resources for replication networks
Array (storage)-based 򐂰 High performance 򐂰 High cost

(specifically DS8000 Copy 򐂰 Operating system 򐂰 Requires knowledge
Services) independent specific to the type of
򐂰 Does not consume host storage platform used for
resources for migration the migration
򐂰 Complex management
򐂰 Might require a quiesce of
the database

Migration technique Pros Cons
Appliance-based (SAN Volume 򐂰 Migration can happen 򐂰 Requires an initial outage to

Controller) online during peak hours bring the host volumes
򐂰 Supports heterogeneous online to the SAN Volume
environments (servers and Controller
storage) 򐂰 Requires a host reboot to
򐂰 Single point of load or upgrade the
management for migration SDD+SVC device drivers
򐂰 No System z support with
SAN Volume Controller
Tape backup/restore-based 򐂰 Does not require additional 򐂰 Requires the disruption of

(Legato, Tivoli Storage special tools, software, and the applications and
Manager, Net-backup, and so hardware downtime
on) 򐂰 Does not require additional 򐂰 Slow and cumbersome
skills or training
2.2 Key decision factors

During the planning, you need to determine what hardware or software is needed to
successfully perform the migration. Here are key factors to consider and a brief description
about how the products highlighted in this book address that specific factor.
2.2.1 Migration performance

One consideration is migration performance, which is how quickly data is copied from the
source to the target. You need to balance performance against network bandwidth and
system resources. For example, data that is copied at a high rate but consumes too many
system resources can severely affect production application or system performance. If data is
copied too slowly, the migration process can take a significant amount of time, potentially
prolonging downtime of an application.
Various tools have the following capabilities to pause, pace, or throttle the migration speed:
򐂰 Throttling of host-based data migration rate using LVM or LDM depends on the tool
selection. Certain advanced Volume Managers provide a function that adjusts the data
mirroring synchronization rate. The function either manipulates the number of processes
or threads running in parallel (multithreading), or increases or decreases the delay
between I/O within a single thread.
򐂰 TDMF includes a throttling/pacing capability that can adjust data movement.
򐂰 SAN Volume Controller has the built-in capability to specify the number of threads at which
the data is copied from source to target.
򐂰 IBM Copy Services using Rational Method Composer can also prioritize the host I/O and
Rational Method Composer writes. It does so by using dedicated paths for the Rational
Method Composer channels. No impact to host I/O is experienced.
2.2.2 Migration backout scenarios, protecting source and target volumes

Data migration backout scenarios allow you to roll back the migration. If something goes
wrong, you can terminate the migration, and restart or continue application processing on the
original data set. With tools such as LVM, you can use a function that moves the data from the
source to the target. This approach makes migration backout more complex because it
requires all the data to be moved back to the source. Another approach is to mirror and

synchronize data between the source and the target. This approach allows you to choose
which copy of data to use at the time you disassociate the target from the source. The state of
the source volume remains preserved and the volume can be used independently from the
target volume if anything goes wrong during the migration.
The available tools have the following advantages and disadvantages:

򐂰 TDMF allows you to preserve the state of the source volume. If an error occurs during
migration, TDMF switches back automatically to the source volumes.
򐂰 Host-based utilities such as LVM for UNIX and LDM for Windows can migrate data while
applications are up and running. Choose a Volume Manager mirroring technique that
allows you to break the mirrored copy of a volume into two or more physical copies. Each
physical copy can then be activated, mounted, and used independently. If a copy becomes
unusable, it is more practical to revert to another copy. Activating and mounting file
systems on an independent copy of a volume is the least time consuming way to back out.
򐂰 With SAN Volume Controller, the source volume is first brought under the control of the
SAN Volume Controller as an image mode device. The data remains in its original form on
the source storage platform. If the host does not recognize the volume presented through
the virtualization layer, this step can be undone. The server can then be reconnected to
the source volume directly. After the source volume is converted from image to managed
mode, the source volume can no longer be accessed directly. You must use the SAN
Volume Controller virtualization layer. Converting the volume from image mode to
managed mode using SAN Volume Controller is similar to moving data using LVM. When
the data movement is done, the source volume can no longer be used. There are SAN
Volume Controller tools similar to LVM mirroring where an extra copy of a volume can be
created and used for backout. These tools and techniques are addressed in more detail in
Chapter 7, “SAN Volume Controller-based migration” on page 175.
򐂰 IBM Copy Services using Rational Method Composer allows the migration to be
terminated and restarted from the beginning, or paused and restarted from the last write
to the target disk.
2.2.3 Understand target storage requirements

Before starting the data migration to a target storage platform, you need to plan for optimal
volume allocation at the target. For environments with no storage virtualization capabilities,
data migration presents an opportunity for storage optimization. If planned properly, you can
achieve the following improvements in volume allocation:
򐂰 Performance optimization
򐂰 Resiliency
򐂰 Migration to another storage tier
򐂰 Reclamation of unused capacity
Before the data migration takes place, create a set of target volumes. Thorough planning for
the target volume allocation is essential for storage optimization. Storage optimization is a
complex task and the optimization goals can be contradictory. You might need to prioritize
based on your company standards and available tools and techniques available for the data
migration. Any mistakes made at this stage of the data migration process will be difficult to
correct in an environment with non-virtualized storage.
A virtualized storage environment can have nondisruptive data movement, so storage

optimization that can be performed without affecting applications and users. In a
non-virtualized storage environment, however, the data migration is a disruptive process.
Often the only opportunity for optimization of non-virtualized storage is when the data is
migrated.

Understanding the architecture of a target disk platform is important during the planning
phase. If the architecture is not properly understood, you might create configurations with
suboptimal performance, availability, and scalability characteristics. Understand your key
applications disk access patterns so you can optimize performance and availability
characteristics of the most critical applications. Storage tiering is a common technique used
for workload prioritization. However, even after an application is identified for a specific
storage tier, there are other allocation aspects that need to be considered:
򐂰 Improving the resiliency of a database server
򐂰 Avoid having database volumes and transaction logs on the same RAID group. Careful
placement reduces the risk of losing data as a result of a RAID group failure.
Architecture of various disk platforms, allocation techniques, connectivity, storage tiering, and
alignment with applications are complex tasks. For more information about architectural
considerations for specific disks subsystems, see 2.4, “Preparing DS8000 for data migration”
on page 27.
Migration tools have different capabilities for migrating data to volumes with differing
capacities. Tools such as UNIX LVM-based tools are the most flexible when it comes to
dealing with the differences in volume capacities. Windows LDM migrations require the target
volume to be the same size or larger.
IBM Copy Services-Based migrations in most cases require the target volume to be of the
same capacity as the source. Generally, work with matching capacity volumes. You might
need to reverse the direction of data copying, which is not allowed from a volume of greater
capacity to a lesser one. After the migration is completed, you can freely increase the
capacity of a volume if required. Use operating system tools to recognize the increased
capacity of an underlined volume, and to expand boundaries of a file system on that volume.
SVC-Based migration from image mode to managed mode virtual disk cannot involve any
volume capacity changes. After the migration is finished, the target volume capacity can be
dynamically expanded. This process is similar to SVC-Based data migrations using Copy
Services or volume mirroring techniques.
Important: Be careful that the data capacity on the source volume is not greater than the
free space on the target volume.
2.2.4 Different source and target hardware

Another common migration situation is unlike source and target storage hardware. Although
host-based products allow migration to unlike storage devices, ensure the DS8000 minimum
level requirements are met for array-based migrations. The different tools have the following
considerations:
򐂰 TDMF is hardware-independent, allowing nondisruptive migration in multivendor
environments.
򐂰 SAN Volume Controller provides a virtualization level that isolates the back-end storage
type from the host. It therefore allows any supported storage servers data to be migrated
to any other type of supported storage platform.

򐂰 Host-based LVM /LDM tools allow for unlike source and target platforms, but might be
limited in which device drivers can coexist on the host operating system. Research driver
compatibility with the system administrators for each type of platform and operating
system type.
򐂰 Tape-based products run on just about any type of platform and operating system. Check
with your backup/recovery administrator for information.
2.2.5 Application downtime

One of the primary reasons that data migrations occur during off-hours is to avoid application
downtime during peak periods. Depending on the type of data and applications you are
migrating, you might have only a narrow downtime window. For example, having the HR
system is offline for 24 hours over a weekend would have less business impact than if the
e-commerce system were offline for the same period. Some systems are so critical that a few
hours or even minutes of downtime, even during off-hours, is unacceptable. If you are
performing a significant upgrade of such a system, using offline data migration tools and
techniques is not feasible.
The different tools have the following capabilities:

򐂰 TDMF allows for nondisruptive data migration, so applications can stay online and
continue to process data throughout the migration process. Some non-IBM products
cannot be moved online. For more information, see the TDMF Installation and Reference
Guide at:
http://www-950.ibm.com/services/dms/en/support/tdmf/zos/v5/download520.html
򐂰 SAN Volume Controller requires an initial reboot to load the device drivers on the host. It
also requires the application to be brought offline while the source volume is discovered.
You can minimize this outage by performing these steps in parallel. The rest of the
migration is done online.
򐂰 LVM/LDM does not require any outage and in certain cases can be done online. The only
step of the emigration process that requires an outage is the HBA and multipathing device
driver installation. An outage is not necessary if the current version of the HBA and
multipathing device driver supports the source and the target disks subsystems. In most
cases, however, an upgrade of HBA device driver is required even if data migration is done
between disk subsystem of the same type. Different versions of disk subsystem microcode
levels often require different versions of HBA BIOS and device driver levels to ensure
compatibility and supportability.
򐂰 IBM Copy Services allow for seamless data copying between the source and the target.
The data is instantaneously available after a synchronous Copy Services relationship
connection is stopped. If a relationship uses asynchronous connection, there will be some
delay between last write to the source volume and when the data is fully synchronized.
When the data copying is finished, the server must be disconnected from the source and
reconnected to the target, applications shutdown, and file systems demounted. This
process might require a reboot depending on the levels of the HBA and multipathing
device drivers. Reboot your server after reconnecting to the target disk subsystem to verify
that the migration is done properly and all file systems are online.

2.2.6 Key decision factors summary
Table 2-2 summarizes the factors to consider when migrating data, and what tools address
those concerns.
Table 2-2 Key decision factors to consider for migration
Key factors Description Capability
Performance How quickly can data be copied 򐂰 TDMF

from the source to the target, 򐂰 SAN Volume Controller
balanced against system 򐂰 IBM Copy Services
resources 򐂰 LVM
Primary volume/Source data If something goes wrong, the 򐂰 TDMF

protection migration can be terminated 򐂰 SAN Volume Controller with
and application processing limitations
restarted or continued on the 򐂰 LVM / LDM
source data or device 򐂰 Tape based
򐂰 IBM Copy Services
Implement tiered storage Moving data to a different array 򐂰 TDMF

or to different storage media for 򐂰 SAN Volume Controller
cost performance without 򐂰 LVM / LDM
disruption
Multi-vendor environments Hardware from several vendors 򐂰 TDMF

is in use, which can result in 򐂰 SAN Volume Controller
data migration between 򐂰 LVM/LDM with possible
different vendor disk platforms restrictions
򐂰 Tape-based
Application downtime Applications have different All with limits

levels of business criticality and 򐂰 TDMF
therefore have varying degrees 򐂰 SAN Volume Controller
of acceptable downtime 򐂰 LVM / LDM
򐂰 IBM Copy Services
Note: Features and functions can vary by operating system.
2.3 Data migration process

Data migration is typically a complex process. To minimize the impact of downtime, reduce
the possibility of data loss, and control costs, employ a consistent, reliable, and repeatable
migration methodology.
The migration process includes the following phases:

򐂰 Planning phase
򐂰 Migration phase
򐂰 Validation phase

A typical migration process is shown in Figure 2-1.
Migration Process
Planning phase Migration phase Post migration
phase
Test &
Analyze Plan Execute Validate
Develop
 Identify affected  Define test  Define Future  Validate HW and SW  Run post-
application environment Storage requirements validation test
 Find dependencies  Develop test and Environment  Customize migration  Perform
between verify migration  Create Migration procedures knowledge
application scenarios Plan  Install and configure transfer
 Analize hardware  Develop  Develop Design  Run pre-validation  Communicate
requirements automation scripts Requirements test project
 Analize bandwidth if required  Create Migration  Perform migration information
requirements  Document step-by- Architecture  Verify migration  Create report on
 Develop feasable step procedures  Obtain Software completion migration
migration scenrios  Develop tools and licences statistics
 Develop back out verification  Start Change  Conduct
scenarios procedures Management migration close
process out meeting
Cut over to Migration

new hardware closed
Figure 2-1 A suggested three-phase migration process
2.3.1 Planning phase

A successful data migration always requires substantial time spent on evaluation and
planning. Adequate planning is the critical success factor in any migration project.
The higher the complexity of an environment and the more critical the data, the more critical
migration planning becomes. Careful migration planning identifies potential issues, allowing
you to avoid or mitigate them. Migration planning also identifies which data to migrate first,
which applications must be taken offline, and the internal and external colleagues you need to
communicate with.
Important: Planning is the critical success factor in a migration project.
The planning process must cover the following considerations:

򐂰 Determine the requirements for the migration
򐂰 Assess the current storage environment
򐂰 Identify requirements for the future state of the environment
򐂰 Create a migration plan
򐂰 Document architectural decisions for specific hardware and software components
򐂰 Develop migration procedures
򐂰 Test plans
Successful migration planning involves more than just the IT staff. The business owners of the
applications and data being migrated must also be included. The owners are the best
resource for determining how important an application or set of data is to the business.
Coordinate the migration with the application owners because uncoordinated tasks can cause
problems during the migration. Do not, for example, plan a migration of the financial system

on the same weekend that the finance department is finalizing quarterly numbers. Keep all
affected application owners involved and informed throughout the migration process
Assessment of the environment

Involve the application owner in the early stages of the migration planning process so all
affected applications are considered. Many applications have dependencies on other
applications, some of which might not be obvious. These dependencies can be at any layer of
application stack. For example, a storage device that requires a physical move to another
location affects application A. Because to the resulting bandwidth restriction, the application
server must also be move to the new location. However, Application B must be on the same
IP subnet as application A, so it must be migrated as well.
To uncover and understand the implications of such dependencies, carefully assess the
complete environment using the following steps:
1. Identify all applications that are directly affected by the planned migration.
2. Find application dependencies throughout the entire application stack, all the way down to
the hardware layer.
3. If any additional cross-application dependencies are identified, expand the migration
scope to include the newly identified application.
4. Repeat the previous steps until no additional cross-application dependencies are
uncovered.
5. Identify and document the requirements for a downtime of all the applications in scope.
6. Evaluate the requirements against any restrictions for application downtime.
Additional constraints might be identified during the assessment process, including the need
for extra capacity, connectivity, network bandwidth, and so on. Discovering these additional
requirements early is vital because it gives the project team more time to address them or
develop alternatives.
Test and develop migration scenarios

Usually, more than one set of tools and techniques can be used in a data migration. You need
to evaluate the available tools based on your type of migration and the outcome of the
assessment process. To test these migration scenarios, build a test environment that
duplicates as near as possible the environment on which the migration will take place. If you
cannot construct such an environment, IBM has a service offering that allows you to use IBM
lab equipment. The migration process can then be tested.
When the migration scenarios are developed in the test environment, plan to be able to
backout the migration so you can effectively reset the test environment. When all scenarios
have been developed, tested them extensively until you are confident that the scenarios will
work during the migration of live data.
You might also want to develop automation scripts. The following examples demonstrate
situations where automation scripts are useful:
򐂰 When the time frame of the migration is weeks or months. Changes to the production
environment, like storage expansion, might happen during the migration, and these
additional volumes must be taken into account.
򐂰 When the scenarios are so complex that operators need guidance. These instructions can
be provided with a script. For example, in a migration where the scenarios are issued in a
defined sequence, the operator can be guided step by step using a script.

򐂰 When reliable status verifications are difficult to get, especially in a situation where many
volumes must be migrated. A script automating the status check can reduce errors,
particularly when the volume addresses are not in a contiguous order.
Budget sufficient time and attention for this important phase. The better you test the
scenarios and provide automation, the smoother the live migration will go. The team running
the migration must be familiar with every step of the development process, and must feel
confident with the proposed timelines. Strive to minimize the risk of unplanned outages or
other complications during the migration.
Thorough development and testing of the migration process can reduce the potential impact
of the migration and increasing success rate of individual migrations.
Migration plan
As part of the planning and preparation, create a high-level migration plan (Table 2-3) and
communicate the plan to all stakeholders.
Employ checkpoints to ensure that the migration is proceeding correctly. It can be

overwhelming, for example, to consider migrating several servers in one large move. Instead,
break down the migration into smaller activities to make it more manageable. Among other
benefits, checkpoints simplify any needed backouts. Therefore, create an overall
comprehensive plan and then break individual tasks into subplans.
The plan also allows you to track schedule commitments while completing the migration.
Always allocate extra time for tasks, generally 15 - 20% more time than would be required for
the best case migration scenario. This gives you time to resolve unexpected issues. The
better you plan, the fewer unexpected or unforeseen issues will be encountered with both
resource (team members) commitments and technology.
Table 2-3 Sample high-level migration plan
Action item Assigned to Status Date
1. Establish a migration
management team.
2. Gather availability and

production schedules.
3. Document Change Control

procedures so they can be
incorporated into the migration
procedures and plans.
4. Document the timeline for

activities for both hardware
changes and the data migration.
5. Announce the migration early in

the cycle. Adhere to the Change
Management process to
determine how much advanced
notice needs to be given.
6. Gather information about the

storage server environment and
applications (lists and drawings).
7. Inform the security and

compliance groups about the
migration.

Action item Assigned to Status Date
8. Schedule a pre-migration
rehearsal that includes all the
members on the migration team
and a representative data
sampling.
9. Establish a Migration Status

call-in process.
10. Use a Migration Planning

checklist to make sure all of the
pre-migration planning steps
have been executed.
Detailed information list for the migration plan

This section addresses the migration plan action items in more detail.
1. Establish a migration management team and create a technical migration team.
Some environments have team members who have several roles and responsibilities.
Team members might include, but are not limited to, the following roles:
– Project manager
– Client (account) manager
– DBA/Application owners
– System administrator
– Network administrator
– Security administrator
– Firewall administrator
– Disk storage administrator
– Backup/Recovery administrator
– SAN fabric administrator
– Hardware CE
– Floor planner
– Cable vendor
– Disaster/Recovery administrator
– IT Architect
2. Gather availability and production schedules.
Determine what outage windows are available or need to be scheduled. Some activities
call for multiple outages. Planning outage times will help the overall migration run
smoothly.
Application availability can be limited for the following reasons:
– Month/end /quarterly processes
– IBM FlashCopy® or Metro/Global mirror/copy running processes and their time
restrictions
– Database/application refreshes
3. Document Change Control procedures so they can be incorporated into the migration
procedures and plans.
Educate all team members on using the change control tool. Successful coordination
allows tasks owned by individual team members to be run in parallel or in a timely fashion.
Some change procedures require lead times of seven days or more before they can be
completed. Any emergency changes can also be communicated to the team.

Following the Change Control process helps avoid missing Service Level Agreement
(SLA) and contractual obligations.
4. Document the timeline for activities for both hardware changes and the data migration.
Develop a realistic timeline with built-in buffer times to create realistic client and team
expectations. It can also show if your progress in the overall migration is on schedule,
allowing you to more adjust and re-evaluate the plan as needed.
5. Announce the migration early in the cycle. Use the Change Management process to
determine how much advanced notice is needed.
Announcing early allows you enough time to adjust the plan if something is discovered
that might cause a potential slowdown. It is critical to warn all affected parties before
touching and moving data. Keep application owners informed by publicizing the migration
plan and giving them a chance to identify flaws.
6. Gather information about the storage server environment and applications (lists and
drawings).
You must understand the following design requirements thoroughly:
– Migration and replication requirements,
– Time schedule
– Vendors involved
– Configuration of the hardware.
When sizing data migrations, there are many key items to consider:
– The number of servers
– The operating system levels
– The amount of storage
– The volume managers
– Types of databases and applications
– Network speeds
– Server clusters
When creating the time, create estimates for planning (typically 25% of the total),
installation and setup time, data copy time, and production cutover.
Your specific requirements help determine the best technology to use for your migration.
Make a list of all the components involved in the migration like the example shown in
Table 2-4.
Table 2-4 Design requirements
Action item Server environment
Server manufacturer (Compaq, Dell, IBM, HP, Sun)
Number of processors
Number of logical partitions (LPARs) or domains
Type of file system (UFS, VxFS, HFS, JFS, JFS2 (inline or outline), NFS, NTFS,
FAT, FAT32
Operating system (OS) version (AIX 5.1, zOS 1.4, IBM System i®)
OS addressing (31-bit, 32-bit, 64-bit)
Databases to be moved (DB2, IBM Informix®, Oracle, SQL, Sybase)
Database version
Database size

Action item Server environment
Availability requirements of databases (any existing SLAs and downtime issues to

consider)
System DASD
Cluster environment (MSCS, Veritas, Sun, IBM HACMP™, MC/Service Guard)
Storage environment
Storage vendor and model (EMC, HDS, IBM, STK, Dell, HP)
Channel type (IBM ESCON®, IBM FICON®, Fibre Channel, iSCSI, SAN)
SAN HBA and model (QLogic, Emulex, JNI)
Number of channel paths
Logical- to-physical mapping (that is, RAID-1 versus RAID-5)
Number of source volumes to be migrated
Volume sizes
Identify target volumes to receive source data
Network environment (if applicable)
Topology
Channel extension technology
Speed of network
7. Identify target storage requirements.

Work with your storage administrator, to understand and outline the new storage
configuration and layout. Identify specific requirements for volume RAID type, size, and
disk and spindle isolation versus spread ahead of time. For more information, see 2.4,
“Preparing DS8000 for data migration” on page 27.
8. Inform the security and compliance groups about the migration.
9. Schedule a pre-migration rehearsal that includes all the members on the migration team
and a representative data sampling.
A rehearsal allows you to uncover any flaws in your plan. Ideally, test a complete migration
cycle using test equipment.
10.Establish a Migration Status call-in process.
Having a regular phone meeting during the migration helps keep the migration on
schedule and avoid problems.
11.Use a Migration Planning checklist to make sure that all pre-migration planning steps are
executed.
Table 2-5 illustrates an example checklist for migration planning.
Table 2-5 Migration methodology and approach
Action Migration and validation methodology checklist
item
Structure the migration architecture to match the production requirements.
Use checklists to ensure that operating patches and software are at the correct
levels.

Action Migration and validation methodology checklist
item
Build detailed migration procedures following the chosen architecture.
Put together a schedule of the migration procedures.
Establish a test plan to validate the installation of all required components.
Develop a cooperative deployment plan.
Write and configure any automation scripts.
Run a simple test that validates the migration process.
Implement the migration procedures and timeline built in the design phase.
Verify the migration completion by checking the status of the migration jobs.
Migration phase
During the migration phase, communicate your plans and obtain, install, and configure
hardware, software, automation scripts, and tools needed to perform the actual data
migration. Run a pre-migration data validation test in addition to post-migration validation
testing. For more information about validation, see 2.3.2, “Validation phase” on page 27.
These tests confirm that the data is in the same state after the migration as it was before. Test
your plan on a test or development (non-production) environment if possible.
The most important part of this stage is the actual migration itself. As outlined, proper
methodology can simplify this process by:
򐂰 Enhancing the speed of migration
򐂰 Minimizing or eliminating application downtime
򐂰 Allowing migration during regular business hours
Table 2-6 illustrates a high-level sample plan only. Customize it for your specific environment.
You can use it as a model, or write your own plan. Part 2, “Migration scenarios” on page 39
contains several sample plans for host-based, network-based, array-based, and
appliance-based migration plans. See the sample plan that best matches your migration to
become familiar with the concepts and principles. The details need to be worked out with the
migration team members. This example plan illustrates what steps might be taken to migrate
Windows SAN volumes to a DS8000 using SAN Volume Controller.
Table 2-6 High-level test project plan for a single host server
Location Activity Owner Comment
Validate Change Record is in place Storage Admin

and is approved
Host server/ Fabric Verify the WWPN numbers of the Storage Admin
or Storage unit HBAs with the System Admins and System
Admin
Host server Update SDD drivers System Admin
Host server Reboot host server System Admin
Host server Validate ESS LUNs can still be System Admin

accessed
Host server Power down host server System Admin

Contact Storage Team when System Admin

finished
ESS unit Locate LUNs on ESS unit for this Storage Team
host server
ESS unit Unassign ESS LUNs from host Storage Team

server
ESS unit Assign ESS LUNs to the SAN Storage Team

Volume Controller
SAN Volume Create host definition Storage Team

Controller cluster
SAN Volume Discover Mdisks Storage Team

Controller cluster
SAN Volume Change Mdisk names Storage Team

Controller cluster
SAN Volume Create Image Mode/Unmanaged Storage Team

Controller cluster Vdisks in Mdiskgroup IMAGEGRP
SAN Volume Map Vdisks to host Storage Team

Controller cluster
Contact System Admin team when Storage Team

completed
Host server Power up host server System Admin
Host server Validate Vdisks can be accessed by System Admin

host server
Host server Start applications System Admin
Host server Turn over to customer System Admin
Contact System Admin Team when System Admin

completed
SAN Volume Migrate ESS Vdisks to DS8300 Storage Team

Controller cluster Mdiskgroup
SAN Volume Rename ESS Vdisks Storage Team

Controller cluster
SAN Volume Remove ESS Mdisks from Storage Team

Controller cluster IMAGEGRP Mdiskgroup
ESS unit Unassign ESS LUNs from SAN Storage Team

Volume Controller

ESS unit Remove host definition from ESS Storage Team

unit
Remove the ESS to host zone Storage Team
SAN Volume Validate Mdisks no longer defined Storage Team

Controller cluster on SAN Volume Controller
2.3.2 Validation phase

After migrating data, ensure that the target system contains a complete and accurate copy of
the source. You also need to make sure that the migrated data is still usable by applications.
Note: It is important to validate that you have the same data and functions of the
application after the migration. Make sure that the application runs with the new LUNs, that
performance is still adequate, and that operations and scripts work with the new system.
After you complete the migration, compile migration statistics and prepare a report to highlight
what worked, what did not work, and any lessons learned. Share this report with all members
of the migration team.
These reports are critical in building a repeatable and consistent process by building on what
worked, and fixing or changing what did not. Also, documenting the migration process can
help you train your staff, and simplify or streamline the next migration you do, reducing both
expense and risk.
Make sure during the validation phase that the backout scenarios can still be applied. If there
are problems with the data or the infrastructure after the migration, you can fix the problem or
revert to the original state.
Also, keep the data at the original site available and usable. Maintaining this data allows you
to discard the migrated data and restart production at the original site if something goes
wrong with the migration. Another migration can then be run after you fix the problems without
affecting your applications.
2.4 Preparing DS8000 for data migration

When planning a migration to the DS8000, review the overall concepts and architecture of the
storage subsystem. This review helps you to prepare the logical system configuration for the
databases or data files you are migrating.
Make sure that you understand and document the following items:
1. The logical layout of the Arrays and LUNs (understanding array characteristics with
workloads and data placement on arrays and LUNs)
a. The size of the arrays
b. The number of arrays
c. The type of arrays (6+P) versus (7+P)
d. The array locations in relationship to the owning DA Pair
e. Logical LUN (Volume) sizes versus physical disk sizes

f. The number and size of DS8000 LUNs you create to match the source Storage layout
g. RAID levels
h. Host striping
2. The DS8000 microcode level and compatibility to support the host SDD or multipathing
driver
2.4.1 Available hardware resources

Place data based on the hardware resources available. Determine how many of the following
hardware resources you have:
򐂰 DA pairs
򐂰 Ranks (arrays, array sites)
Tip: You will always see two servers (0 and 1).
򐂰 I/O enclosures
򐂰 Host adapters
򐂰 I/O ports are available on each host adapter
Evenly distribute the data across as many hardware resources as possible. However, you
might want to isolate data to specific hardware resources for guaranteed resource dedication
to that data I/O. For best performance results, create only one rank per extent pool. Limiting
the ranks helps you to map and identify problem resources throughout the lifetime of the data
or database. Performance issues might arise later because of the following changes:
򐂰 Database growth
򐂰 Saturation of certain hardware resources as the database changes
򐂰 New data or files/filesystems being created on the same set of hardware resources,
changing the performance of all data on those resources
2.4.2 Understanding array characteristics

It is beyond the scope of this book to address the performance characteristics of arrays, LUNs
and the logical layout of the DS8000. The following is a basic understanding of these
characteristics. Preparing the DS8000 properly can minimize performance impacts on the
application databases and file systems both during and after the migration.
Remember: The following explanation applies only to the initial creation of LUNs in the
array.
Figure 2-2 on page 29 shows an example of how the data workloads on the disk arrays affect
one another. The example is a DS8000 array of eight DDMs, in a 7+P format with no spare.
The eight physical disks are divided into 5 logical LUNs.
The first LUN (logical-disk1) is formed from strips of sections (numbered 1) along the outer
edge of each of the DDMs making up the array. Subsequent LUNs are similarly formed from
areas on the DDMs, each sequentially closer to the center.

Figure 2-2 How LUNs are created in an array
Tip: This example is only true if there is a one-to-one relationship between rank and extent
pool. If an extent pool has multiple ranks, the LUN might span ranks.
LUNs from the same array can be assigned to the same or separate servers on an individual
basis. Workload sharing or isolation depends on the LUN to array mapping. An example of
two databases or workloads on the same array is shown in Figure 2-3.
Figure 2-3 How databases share an array
A database might consist of two of the LUNs (logical volumes made up of strips 1 and 3) in
the array. Another database might consist of the other three logical volumes (logical disks 2,
4, and 5) in the array shown in Figure 2-3. The volumes in the array can even be assigned to
two different servers.
Because the disks share physical heads and spindles, I/O contention can result if the two
application workloads peak at the same time. Prepare the logical configuration of arrays,
ranks, extent pools, and LUNs for the DS8000. You must meet or exceed the I/O throughput
parameters of the source storage platform you are migrating from.
Larger arrays always outperform smaller arrays. For example, an RAID-5 (7+P) array
outperforms a (6+P) array. Data that needs to perform better would do better on a 7+P array
when using RAID-5 and (4x4), rather than (3x3) when using RAID-10 arrays.
Important: The speed of the drives is also a factor to consider. An array made up of 15-K
DDMs outperforms one consisting of 10-K DDMs. Speed is an important consideration
when moving to a DS8000 with mixed speed DDMs.
The 7+P or 4x4 array outperforms a 6+P or 3x3 array because the array will stripe data
across more disks. For example, writing data across a 7+P RAID-5 array will stripe across

eight disks (Figure 2-2 on page 29) rather than seven disks (Figure 2-4). A 4x4 RAID-10 array
writes the same data twice striped across four disks rather than three disks.
Chunk1 Chunk1 Chunk1 Chunk1 Chunk1 Chunk1 Parity
Chunk2 Chunk2 Chunk2 Chunk2 Chunk2 Parity Chunk2
Spare
Chunk3 Chunk3 Chunk3 Chunk3 Parity Chunk3 Chunk3
Chunk4 Chunk4 Chunk4 Parity Chunk4 Chunk4 Chunk4
Chunk5 Chunk5 Parity Chunk5 Chunk5 Chunk5 Chunk5
Chunk6 Parity Chunk6 Chunk6 Chunk6 Chunk6 Chunk6
Parity Chunk7 Chunk7 Chunk7 Chunk7 Chunk7 Chunk7
Figure 2-4 A 6+P Array with a spare
Generally, however, try to balance workload activity evenly across RAID arrays, regardless of
the size. The cache mitigates most of the performance differences, but keep in mind this
guideline when you are fine-tuning the DS8000 I/O throughput.

3
Chapter 3. IBM service offerings

IBM Data Migration Services, part of IBM Storage and Data Product Services, can provide
resources, technology, and expertise. IBM Data Migration Services can help you simplify your
data migration process in open systems and mainframe environments.
The services combine the experience of IBM services professionals with the following
powerful migration tools:
򐂰 IBM System Storage SAN Volume Controller
򐂰 Transparent Data Migration Facility (TDMF)
򐂰 z/OS Dataset Mobility Facility (zDMF)
IBM can help with project planning and management, and can provide technical assistance
for migrating data including the following activities:
򐂰 Hardware environment refreshes
򐂰 Storage reclamation
򐂰 Consolidation
򐂰 Data center migrations
Data Migration Services are one part of the IBM Global Services portfolio of Data Center
Services. Data Center Services includes Storage Optimization and Integration Services for
end to end storage consulting.
The following data migration services from IBM Global Services are currently available:
򐂰 IBM Migration Services for data for Open Systems
򐂰 IBM Migration Service for System z Data
򐂰 IBM Migration Services for Network Attached Storage Systems
In partnership with IBM Global Services, IBM Systems and Technology Group offers services
for early adopters of IBM technology and the following migration services:
򐂰 IBM XIV Migration Services
򐂰 IBM DS8000 Data Migration Services using temporary Licenses for Copy Services
This chapter includes the following sections:

򐂰 IBM Global Services offerings
򐂰 IBM Systems and Technology Group Lab Services

3.1 IBM Global Services offerings
The following services are provided by IBM Global Services divisions around the world:
򐂰 IBM Migration Services for data for Open Systems
򐂰 IBM Migration Services for System z data
򐂰 IBM Migration Services for Network Attached Storage Systems
3.1.1 IBM Migration Services for data for Open Systems

IBM Migration Services for open systems attached to System Storage disk systems provides
services performed by a technical specialist to plan and manage nondisruptive migration.
Migration is to IBM System Storage Technologies from servers running the following
operating systems:
򐂰 IBM AIX
򐂰 HP-UX
򐂰 Sun Microsystems Solaris
򐂰 Microsoft Windows technology
This migration can be accomplished with minimal and often no interruption to service using
mirroring. Mirroring is an operating system or software/hardware tool. It allows your data to
be replicated to the IBM System Storage disk system or other storage vendor products. In
addition, IBM can provide a migration control book that details the activities performed during
delivery of the services.
In executing these projects, IBM uses the following technologies and products, among others:
򐂰 IBM System Storage SAN Volume Controller
򐂰 OS-Specific mirroring
򐂰 Global Copy
򐂰 Softek Transparent Data Migration Facility (TDMF)
Value proposition
Using IBM Migration Services can help you achieve the following goals:
򐂰 Reduce or eliminate downtime and data loss
򐂰 Preserve data updates throughout the migration, allowing the process to be interrupted if
necessary
򐂰 Improve post-migration management (a migration control book is provided that explains
the work performed)
Benefits
The implemented solution allows you to realize the following benefits:
򐂰 Professional, speedy, and efficient planning and implementation of data migration
򐂰 Greater flexibility and improved data migration capabilities
򐂰 Skills instruction for members of your staff
򐂰 Focus on business-critical activities
򐂰 Opportunities to reduce the total cost of your IT infrastructure

Key questions and considerations
Use following questions to determine whether IBM Migration Services are right for you:
򐂰 Do you have a requirement to efficiently migrate your data?
򐂰 Do you need assistance in the planning and execution of data migration?
򐂰 Do you have the available resources and technical expertise within your enterprise to
manage your data migration?
򐂰 Does your staff have the necessary skills to perform this migration?
򐂰 Can you afford to have outages during the process?
Deal size/pricing
Engagement pricing varies depends on the following factors:
򐂰 Amount of data being migrated
򐂰 Complexity and type of data being migrated
򐂰 Migration method and technology
򐂰 Travel and living expenses
Learn more
IBM TotalStorage hardware-assisted data migration services are available around the world.
For more information, visit the following web address:
http://www-935.ibm.com/services/us/en/it-services/data-migration-services.html
3.1.2 IBM Migration Services for System z data

IBM Migration Services for System z data provides professional services performed by
technical specialists to help plan and manage the nondisruptive migration of data. Use IBM
Migration Services for System z data if you want to migrate data to DS8000 or other IBM disk
systems from System z attached disk systems.
In addition to providing support for moving data to IBM System Storage products, IBM can
also help you move data among disk systems from other storage manufacturers.
Migration can be accomplished using hardware or software that allows direct access to
storage device volumes to be copied to new storage devices. The migration takes place
without interruption to data availability. IBM can work with your personnel to plan the data
migration activities, and can install the migration software or migration hardware tools in your
environment.
At the completion of these services, data can be transferred from your existing 3380/3390
formatted DASD volumes to the TotalStorage disk system.
In executing these projects, IBM uses technologies and products such as:
򐂰 Softek Transparent Data Migration Facility for z/OS (TDMF z/OS)
򐂰 Softek z/OS Dataset Mobility Facility (zDMF).
򐂰 Global Copy
Value proposition
Using IBM Migration Services can help you achieve the following goals:
򐂰 Reduce or eliminate downtime
򐂰 Preserve data throughout the migration
Chapter 3. IBM service offerings 33

򐂰 Allows movement to and from 3990-compatible storage systems without special
configuration changes
򐂰 Allows movement to and from IBM disk systems, and systems from other leading
manufacturers
򐂰 Combine or split subsystems
򐂰 Move portions of data in an existing storage subsystem
򐂰 Protect data by preserving data updates throughout the migration, allowing the process to
be interrupted if necessary
򐂰 Improve post-migration management (a migration control book is provided that explains
the work performed)
Benefits
򐂰 Professional, speedy, and efficient planning and implementation of data migration
򐂰 Greater flexibility and improved data migration capabilities
򐂰 Skills instruction for members of your staff
򐂰 Focus on business-critical activities
򐂰 Opportunities to reduce the total cost of your IT infrastructure
Key questions and considerations

򐂰 Do you need to migrate individual volumes (not subsystems) so you can migrate multiple
volumes at the same time?
򐂰 Do you need to combine or split subsystems, or move portions of data in an existing data
storage?
Deal size and pricing

Engagement pricing varies depending on the following factors:
Learn more
IBM TotalStorage hardware-assisted data migration services are available around the world.
http://www-935.ibm.com/services/us/en/it-services/migration-services-for-system-z-
data.html

3.1.3 IBM Migration Services for Network Attached Storage Systems
IBM Migration Services for Network Attached Storage provides migration services to help you
deploy IBM network-attached storage products. IBM offers specialists who can help you with
planning and managing the entire data migration process. They will address issues related to
complexity, disruption, system performance, hosting, availability, and data integrity.
Value Proposition
IBM Migration Services for Network Attached Storage combines proven migration methods
and tools with the planning and management experience of highly skilled IBM storage
specialists. IBM offers you continuous access to critical data throughout the migration process
to mitigate project risk. IBM also uses new storage and data platforms for the best IT and
business results. Additionally, this service provides both hardware- and software-based
migration options that are customized to your needs.
Benefits
򐂰 Increased value of investment in network-attached storage because of faster migration
򐂰 Reduced system downtime
򐂰 Reduced risk associated with data transfers
Key Questions and Considerations

򐂰 Do you need to migrate individual volumes (not subsystems) so you can migrate multiple
volumes at the same time?
򐂰 Do you need to combine or split subsystems, or move portions of data in an existing data
storage?

Engagement pricing varies depending on the following factors:
Learn More
IBM Migration Services for Network Attached Storage (NAS) are available around the world.
http://www-935.ibm.com/services/us/en/it-services/storage-and-data-services.html

3.2 IBM Systems and Technology Group Lab Services
IBM Systems and Technology Group (STG) Lab Services can help you optimize your data
center and system solutions.
STG Lab Services has the knowledge and skills to support your entire information technology
solution. STG Lab Services is focused on the delivery of new technologies and niche
offerings. It collaborates with IBM Global Services and IBM Business Partners to provide
complete solutions to keep your business competitive.
3.2.1 IBM XIV Migration Services

IBM XIV Migration Services from STG Lab Services can help you achieve an efficient and
smooth migration of data to the XIV Storage System with minimal disruption. These services
assist you with the planning, implementation, and validation of data migration from existing
supported hosts and storage subsystems to the new XIV controller.
XIV Migration Services performs the following tasks:

򐂰 Conduct an on-site planning meeting to address optimizing migration and define a sample
migration exercise for a test or production environment
򐂰 Determine personnel skills and other resources required
򐂰 Establish your business requirements including performance, capacity, availability, and
identification of specific data to be migrated
򐂰 Identify security options and considerations
򐂰 Assess your migration readiness
򐂰 Schedule migration activities, including start and end dates
򐂰 Perform migration activities on supported host systems
򐂰 Perform post migration verification on a sample set of data
򐂰 Conduct informal skills transfer throughout the migration activities
򐂰 Update the installation record with details about LUNs migrated
򐂰 Provide basic migration skills instruction for up to three (3) designated technical personnel
Value Proposition
IBM XIV Migration Services from STG Lab Services assists you with data migration to an IBM
XIV Storage System quickly and with minimal disruption. Services include planning,
implementation, validation of data migration, and skills transfer.
Benefits
Using IBM XIV Migration Services from STG Lab Services provides these benefits:
򐂰 Correct migration of data to the XIV Storage System
򐂰 Shorter migration schedule
򐂰 Expert storage architects
򐂰 Reduced risk of disruption to business

Use following questions to determine whether IBM XIV Migration Services are right for you:
򐂰 Do you have the available skills to effectively migrate data to the IBM XIV Storage
System?

򐂰 If you have the skills internally, would it be a better use of your resources to have IBM
provide XIV migration services?
򐂰 Would you like to increase the XIV migration skills of your staff?
򐂰 Do you want to have your data migrated to your new IBM XIV Storage System by highly
skilled and experienced specialists?

Contact an STG Lab Services Opportunity Manager for an estimate of pricing for this service.
Scope and pricing depend on the amount of data to be migrated and the effort required. Data
migration from NAS, AIX Virtual I/O (AIX VIO), or SAN Volume Controller environments
requires special migration methods and must be identified and scoped.
3.2.2 IBM DS8000 Data Migration Services

DS8000 Data Migration Services migrate data from IBM ESS 800 to DS8000, or DS8000 to
DS8000 storage systems. They use temporary licenses for copy services, allowing you to
migrate projects without long-term license commitments.
Value Proposition
When you do not need long-term copy services, IBM DS8000 Data Migration Services allows
you to use Copy Services functions for short-term migration scenarios. It allows enhanced
migrations of data between disk subsystems without the long-term commitment and cost of
licenses.
Benefits
Using IBM DS8000 Data Migration Services provides these benefits:
򐂰 Correct migration of data to the DS8000 Storage System
򐂰 Shorter migration schedule
򐂰 Expert storage architects
򐂰 Reduced risk of disruption to business

Use following questions to determine whether IBM DS8000 Data Migration Services are right
for you:
򐂰 Do you have the available skills to effectively migrate data to the IBM DS8000 Storage
System?
򐂰 If you have the skills internally, would it be a better use of your resources to have IBM
provide DS8000 migration services?
򐂰 Would you like to build DS8000 migration skills on your staff?
򐂰 Do you want to have your data migrated to your new IBM DS8000 Storage System by
highly skilled and experienced specialists?

Contact an STG Lab Services Opportunity Manager for an estimate of pricing for this service.
Scope and pricing dependent on the amount of data to be migrated and the effort required.

3.2.3 Additional Services & Offerings
Additional Services and Offerings are available from STG Lab Services and details can be
found by visiting the following web address:
http://www.ibm.com/systems/services/labservices/
Use the Contact Now link in the right corner to contact STG Lab Services. You will be directed
to fill out a form.

Part 2
Part 2 Migration scenarios

This part provides detailed explanations and illustrations of various disk migration techniques
and products.
This part contains the following chapters:

򐂰 Data migration using IBM Remote Mirror and Copy
򐂰 DSCLIbroker
򐂰 IBM XIV data migration
򐂰 SAN Volume Controller-based migration
򐂰 Using mirroring techniques
򐂰 Using TDMF for z/OS
򐂰 z/OS Dataset Mobility Facility
򐂰 Using TDMF TCP/IP for z/OS

4
Chapter 4. Data migration using IBM

Remote Mirror and Copy
This chapter addresses the IBM System Storage DS8000 Remote Mirror and Copy methods
used for data migration.
This chapter contains the following sections:

򐂰 IBM System Storage Remote Mirror and Copy overview
򐂰 DS8000 user interface for Copy Services
򐂰 Data migration from Enterprise Storage Server Model 800 to DS8000
򐂰 Data migration from DS8000 to DS8000
򐂰 Post-migration tasks

4.1 IBM System Storage Remote Mirror and Copy overview
The Copy Services functions of the DS8000 are a set of flexible data mirroring solutions that
allow replication between volumes of disk storage systems. These functions are used to
implement remote data backup and disaster recovery solutions.
The Remote Mirror and Copy functions are optional licensed functions of the DS8000 that
include:
򐂰 Metro Mirror
򐂰 Global Copy
򐂰 Global Mirror
򐂰 Metro/Global Mirror
In addition, System z users can use the DS8000 for:

򐂰 z/OS Global Mirror
򐂰 z/OS Metro/Global Mirror
For more information about these topics, see the following books:
򐂰 IBM System Storage DS8000: Copy Services in Open Environments, SG24-6788
򐂰 DS8000 Copy Services for IBM System z, SG24-6787
This chapter specifically deals with two members of this family of products, Metro Mirror and
Global Copy.
Exception: The Global Mirror member of the Remote Mirror and Copy family is not
addressed because using IBM Global Mirror is not a practical approach for data migration.
Global Mirror is intended to be a long-distance business continuity solution. The cost and
setup complexity of IBM Global Mirror are not typically justified for a one-time data
migration project.
Remote Mirror and Copy functions can be used to migrate data between IBM enterprise
storage subsystems. These subsystems can include IBM Enterprise Storage Server® Model
800 and Model 750.
Important: IBM Enterprise Storage Server Model F20 is not supported for Remote Mirror
and Copy functions to the DS8000. If you intend to migrate from the Model F20, you must
move the data to a Model 800 or 750 before migrating to the DS8000.
Remote Mirror and Copy migration methods replicate data between IBM storage subsystems,
and are sometimes called hardware-based replication. Remote Mirror and Copy migration
offers the following advantages:
򐂰 High performance
򐂰 Operating system-independent
򐂰 Does not consume application host resources
Global Copy migration methods require you to stop application updates and changes while
the remaining data is moved. Both Global Copy and Metro Mirror require the application host
operating system to acquire the new storage subsystem volumes. The following factors must
be taken into account when planning the data migration:
򐂰 Acquire new migration volumes and make them known to the operating system
򐂰 Modify application configuration files
򐂰 Perform data integrity checks

The steps for performing the data migration using Metro Mirror are similar to the steps
required for Global Copy. For more information, see 4.3, “Data migration from Enterprise
Storage Server Model 800 to DS8000” on page 51. Any differences required for each method
are noted.
Appropriate license codes must be purchased and applied to each of the systems in the
migration scenario to use the Remote Mirror and Copy functions. For information about
licensing requirements, speak to your IBM storage marketing representative.
4.1.1 Metro Mirror

Metro Mirror, previously known as Synchronous Peer-to-Peer Remote Copy (PPRC), provides
real-time mirroring of logical volumes between two supported storage subsystems (DS8000
and Enterprise Storage Server). These subsystems can up to 300 km apart from each other.
It is a synchronous copy solution. Write operations must be completed on both local and
remote copies before they are considered to be complete. It is typically used for applications
that cannot suffer any data loss in the event of a failure. It can also be used effectively as a
data migration tool. Figure 4-1 shows the sequence of a write update with Metro Mirror.
Server 4
write Write acknowledge
1 Write to secondary
2
LUN or
LUN or 3 LUN
LUN oror
volume
volume volume
volume
Write complete
Primary Secondary
(source)
acknowledgment (target)
Figure 4-1 Metro Mirror write sequence
When the application performs a write update operation to a source volume, the following
steps occur:
1. Write to source volume (storage unit cache and nonvolatile storage)
2. Write to target volume (storage unit cache and nonvolatile storage)
3. Signal write complete from the remote target DS8000
4. Post I/O complete to host server
The Fibre Channel connection between the local and the remote disk subsystems can be
through a switch. Other supported distance solutions include dense wave division
multiplexing (DWDM).
Chapter 4. Data migration using IBM Remote Mirror and Copy 43

Metro Mirror performance considerations
Because Metro Mirror is a synchronous mirroring technology, it decreases performance for
the write operations. However, unlike operating system mirroring, Metro Mirror does not
consume any processor activity. Use bandwidth analysis and capacity planning to determine
how many Metro Mirror links you need to ensure the best possible performance.
As part of your implementation project, identify and distribute hot spots across your
configuration to manage and balance the load.
Keep in mind the following basic considerations:

򐂰 Is your Fibre Channel network bandwidth between storage subsystems too small, causing
an increase in the response time of your applications at moments of high workload?
򐂰 Do not share the Metro Mirror link I/O ports with host attachment ports. Sharing ports can
result in unpredictable performance of Metro Mirror, and make performance problem
analysis more complicated.
򐂰 Distance is an important topic. Data writes must go to the secondary disk subsystem, and
then be acknowledged. With latency times of active components on the way, plan for
approximately a 1 ms delay per 100 km for write I/O.
򐂰 Sometimes the problem is not Metro Mirror, but rather hot spots on the disks. Be sure that
these problems are resolved before you start with Metro Mirror. Invest the time necessary
to plan the logical configuration of the target DS8000 to lay out the data. Be sure to plan
for high I/O rates and future growth.
4.1.2 Global Copy

Global Copy, previously known as Peer-to-Peer Remote Copy Extended Distance
(PPRC-XD), copies data non-synchronously for both open and z/OS systems. Global Copy is
a great tool for data migrations, especially for longer distances than are supported with Metro
Mirror.
When operating in Global Copy mode, the source volume sends a periodic, incremental copy
of updated tracks to the target volume instead constant updates. Incremental updates cause
less impact to application writes for source volumes and less demand for bandwidth
resources, allowing a more flexible use of the available bandwidth.
With Global Copy, write operations complete on the source disk subsystem before they are
received by the target disk subsystem. This capability is designed to prevent the local
performance from being affected by wait time of writes on the remote system. Therefore, the
source and target copies to be separated by any distance.

Figure 4-2 illustrates the write sequence for Global Copy.
Server 2 Write acknowledge

write
1 3
LUN or LUN
LUN oror
LUN or
volume
4
volume volume
volume
Write to secondary
Primary (non-synchronously) Secondary
(source) (target)
Figure 4-2 Global Copy write sequence
The numbers in the figure refer to the following workflow:

1. The host server makes a write I/O to the source (local) storage unit. The write is staged
through cache and nonvolatile storage.
2. The write returns as completed to the host server’s application.
3. The source storage unit sends the necessary data so that the updates are reflected on the
target (remote) volumes. This is a non-synchronous process, so there is a delay. The
updates are grouped in batches for efficient transmission.
4. The target returns write complete to the source storage unit when the updates are secured
in the target DS8000 cache and nonvolatile storage. The source then resets its Global
Copy change recording information.
Data Migration using Global Copy requires a consistent copy of the data to move applications
and servers to the remote subsystem. By design, the data on the remote system is a fuzzy
copy, with the volume pairs in a copy pending state. To create a consistent copy of the
migrated data, the application must be quiesced and the volume pairs suspended. If the pairs
are terminated before quiescing I/O and suspending the pairs, you might lose ordered
transactions to the remote site.
There are two ways to ensure data consistency during migration using Global Copy:
򐂰 Shut down all applications at the primary site and allow the out-of-sync sectors drain
completely.
򐂰 Issue the go to sync command on the Global Copy relationships when the out-of-sync
sectors are approaching or at zero. When the out-of-sync sectors are fully drained, the
pairs are in full duplex mode and there is a consistent relationship. Using go to sync is the
more reliable method.

Global Copy performance considerations
Global Copy employs a two-phase process of transferring data from primary to auxiliary
storage subsystems. The first phase transfers all existing data to an out of sync bitmap, and
this data is transferred to the secondary. When this operation is complete, the volumes are
flagged as first pass true. After the first pass, Global Copy looks at all the tracks that have
changed since the last time it looked in a round robin fashion. Global Copy then moves this
data into the out of sync bitmap, and transfers the new changes to the secondary disk
subsystem.
Global Copy performance depends on the bandwidth of the Fibre Channel interface (the pipe)
between the Global Copy primary and remote storage subsystems. During the first pass,
Global Copy is of no practical use to data migration because the first pass must complete. If
the data to be transferred is 5 TB and the WAN/DWDM interface operates at 27 MBps, allow
about 60 hours for the first pass. This large amount of time is not a reflection on the
performance of Global Copy. Instead, it reflects how long it takes to move 5 TB at 27 MBps.
The pipe must also be large enough to handle the write load of the migration volumes. Using
the same pipe bandwidth previously mentioned, the amount of data being written to all the
migration volumes should be less than 27 MBps. Otherwise, Global Copy will not be able to
deliver the data to the remote storage quickly enough to keep up. As a result, the data at the
remote site will fall further and further behind the data at the primary site. Again, this limitation
is not a reflection on the capabilities of Global Copy. Make sure that the pipe is large enough
to handle the write load of the migration candidate application.
Dedicate the host adapter ports used for Global Copy to the Global Copy migration.
Host-based activities should not use the same ports as Global Copy because this would
negatively affect Global Copy performance.
4.1.3 Target DS8000 configuration considerations

All the normal considerations for ensuring optimum performance of a new application need to
be followed when considering the performance of a migrated application. For example,
consider an existing Oracle application on a Enterprise Storage Server 800 that is coming off
lease. In this example, the application is running well on 16 RAID-5 arrays composed of 15 K
RPM 18 GB DDMs (1.68 TB). Once a week, the application is Flash Copied to another set of
16 identical RAID-5 arrays (another 1.68 TB). This application cannot be successfully
deployed to a DS8000 with two 10 K RPM 300 GB RAID-5 Arrays (3.4 TB).
Although Global Copy would be able to successfully migrate the data, the application
performance after the migration would surely be far less than on the DS8000. This is limited
because the performance characteristics (I/O per second and MBps) of the 16 Enterprise
Storage Server arrays far outweigh the performance capabilities of two DS8000 arrays.
The performance of a migrated application must be considered as though it were a new

application. Use tools such as Disk Magic to properly size the storage. Also, use performance
disciplines like good data layout for I/O balance across multiple arrays, device adapters, and
host adapters. Also take into account considerations for FlashCopy performance.

4.1.4 Automation software: Tivoli Storage Productivity Center for Replication
The IBM Tivoli Storage Productivity Center for Replication is a useful tool for managing Copy
Services products for the Enterprise Storage Server and DS8000. These products include
Global Mirror, Global Copy, and FlashCopy. It also manages FlashCopy and Metro Mirror for
the SAN Volume Controller. Tivoli Storage Productivity Center for Replication manages Copy
Services in the following ways:
򐂰 Automating administration and configuration of these services with wizard-based sessions
and copy set definitions
򐂰 Allowing control of copy services tasks, including starting, suspending, and resuming
򐂰 Offering tools for monitoring and managing copy sessions
If you are using Tivoli Storage Productivity Center for Replication to manage Copy Services,
use it to facilitate data migration when using Copy Services as your migration technique.
If you are not using IBM Copy Services, do not use Tivoli Storage Productivity Center for
Replication because it adds unnecessary cost and complexity to the implementation. For a
one-time migration project using Copy Services, use Global Copy controlled by one of the
DS8000 user interfaces.
4.2 DS8000 user interface for Copy Services

There are two native user interface methods for the DS8000: IBM System Storage DS8000
Storage Manager graphical user interface (DS GUI), and the DS Command Line Interface
(DS8000 CLI). Both of these user interfaces can be used to start copy services functions.
Additionally, System z Interfaces with TSO and ICKDSF can be used to start the DS8000
copy services functions.
4.2.1 IBM System Storage DS8000 Storage Manager

The IBM System Storage DS8000 Storage Manager (DS Storage Manager) software is
installed on the Hardware Management Console (HMC) provided with the DS8000. The
software can also be installed on a separate personal computer running Windows or Linux.
The DS Storage Manager allows you to control Copy Services functions in real time. For
specific information about installing on a personal computer, see the IBM System Storage
DS8000 User’s Guide, SC26-7915.
The DS Storage Manager can be accessed through the Tivoli Storage Productivity Center
Element Manager of the System Storage Productivity Center (SSPC). It can be access from
any network-connected workstation with a supported browser. It can also be accessed
directly from the DS8000 management console using the browser on the HMC.
Remotely connect to the DS Storage Manager using the IP address or fully qualified name
resolved by the DNS server. The correct port number is the address as shown in the following
example.
https://10.0.0.1:8452/DS8000/Login
In this example, 10.0.0.1 is the IP address of the HMC. 8452 is the port number, and
DS8000/Login is required to access the Storage Manager.

You are prompted for a user name and password (typically managed by the System
Administrator) as shown in Figure 4-3.
Figure 4-3 DS8000 Storage Manager login window
The login window for newer versions of DS8000 is shown in Figure 4-4.
Figure 4-4 DS8000 new Storage Manager login window from release 6.1

For specific details about using the Storage Manager for data migration, see 4.3, “Data
migration from Enterprise Storage Server Model 800 to DS8000” on page 51.
DS Storage Manager is currently supported by the following browsers:

򐂰 Internet Explorer 7.x and 8.x
򐂰 Mozilla Firefox 3.5, and 3.6
4.2.2 IBM System Storage DS8000 command-line interface

The DS8000 command-line interface (DS8000 CLI) provides a full set of commands to
perform all copy services functions. The DS8000 CLI code is installed on one of the
supported open systems platforms (AIX, Windows, Solaris, Linux, or HP-UX). Although
DS8000 CLI is not installed on a System z server, it can create and manage System z
volumes in a FlashCopy, and Remote Mirror and Copy environment. The DS8000 CLI can
also create volumes for all supported open systems servers. Use the latest version of
DS8000 CLI for data migration because it is compatible with earlier versions of the DS8000.
For information about how to create login profiles, see IBM System Storage DS8000
Command Line Interface User’s Guide, SC26-7916. Access can also be obtained by starting
the DS8000 CLI using the TCP/IP address or the fully qualified name of the storage unit
HMC. The user ID and password used for DS8000 CLI are the same as those used for the DS
Storage Manager. For more details about using DS8000 CLI for Remote Mirror and Copy
functions, see 4.4, “Data migration from DS8000 to DS8000” on page 81.
4.2.3 System z interfaces

Several System z interfaces can be used to manage the IBM System Storage DS8000 Copy
Services functions in addition to the Storage Manager and DS8000 CLI. This section covers
the following z/OS options:
򐂰 TSO
򐂰 ICKDSF
These interfaces send commands directly to the DS8000 storage unit over a FICON or
ESCON channel to a conduit count key data (CKD) volume. The command is passed to the
microcode for execution from this volume. The commands are issued the same way for
DS8000 and Enterprise Storage Server.
TSO for Metro Mirror and Global Copy management

For z/OS, the TSO Metro Mirror and Global Copy commands offer a powerful and flexible
interface to manage the Remote Mirror and Copy environment. TSO commands can also be
used for control of open system volumes in a Metro Mirror environment. TSO commands
require at least one CKD volume on the storage units to act as a conduit. TSO commands
communicate with the DS8000 and Enterprise Storage Server through a device number
specified with the command. IP connectivity is not required because the conduit is a FICON
or ESCON channel. TSO commands can be integrated into REXX programs for automation
purposes. Table 4-1 lists the available commands and descriptions.
Table 4-1 TSO Remote Mirror and Copy commands

Command Description
CESTPAIR Establishes Metro Mirror and Global Copy

volume pairs
CESTPATH Establishes Remote Mirror and Copy paths

Command Description
CDELPAIR Deletes volume pairs
CDELPATH Deletes Remote Mirror and Copy paths
CGROUP Controls volume groups with FREEZE and RUN
CQUERY Queries the status of the volumes and paths
CRECOVER Recovers data on the recovery system
CSUSPEND Suspends Metro Mirror and Global Copy volume

pairs
For a detailed description of these commands, see z/OS DFSMS Advanced Copy Services,
SC35-0428, and DS8000 Copy Services for IBM System z, SG24-6787.
ICKDSF for Metro Mirror and Global Copy management

In System z environments, the ICKDSF utility offers a means of control for Metro Mirror and
Global Copy functions. ICKDSF typically runs as a batch program, and can be automatically
run from batch scheduling products. It also supports VM and VSE systems. All commands
must be addressed to a device that is either online or offline to the system where the batch
job is submitted.
The ICKDSF commands used for Metro Mirror and Global Copy management are shown in
Table 4-2.
Table 4-2 ICKDSF Remote Mirror and Copy commands

Command Description
PPRCOPY ESTPATH Establishes Metro Mirror and Global Copy paths

between a primary and secondary LSS
PPRCOPY ESTPAIR Establishes Metro Mirror and Global Copy

volume pairs
PPRCOPY DELPATH Deletes Metro Mirror and Global Copy paths

between a primary and secondary LSS
PPRCOPY RECOVER Allows a system to regain control of a volume on

the secondary DS8000
PPRCOPY SUSPEND Puts a Metro Mirror or Global Copy pair in the

suspended state
PPRCOPY FREEZE Suspends all Metro Mirror or Global Copy

operations at the LSS level
PPRCOPY RUN Resume I/O operations after a freeze with

Extended Long Busy
PPRCOPY QUERY Queries the Meteor Mirror or Global Copy volume

pair and path status
For more information, see the Device Support Facilities User’s Guide and Reference,
GC35-0033.

4.3 Data migration from Enterprise Storage Server Model 800 to
DS8000
Data migration from Enterprise Storage Server to DS8000 using Metro Mirror or Global Copy
requires the following codes:
򐂰 PPRC V2 feature code on the Enterprise Storage Server
򐂰 Remote Mirror and Copy license codes on the DS8000
This section highlights the steps required to set up the configuration to migrate data from
Enterprise Storage Server to DS8000. It includes paths and pairs using DS Storage Manager,
DS8000 CLI, TSO, and ICKDSF for Metro Mirror, and Global Copy. It also includes the
specific steps to complete the data migration from an open systems or a System z
perspective. The final actions to attach the host to the target DS8000 and start the
applications are covered in 4.5, “Post-migration tasks” on page 90.
The following steps are covered:

򐂰 Adding Enterprise Storage Server Copy Services Domain to DS8000
򐂰 Creating Remote Mirror and Copy paths
򐂰 Creating Remote Mirror and Copy pairs
򐂰 Completing the data migration
To establish PPRC pairs, the Enterprise Storage Server must have a PPRC license and the
DS8000 must have a Remote Mirror Copy license. The Enterprise Storage Server must also
have Fibre Channel adapters that have connectivity to the DS8000.
Important: To manage the Enterprise Storage Server Copy Services from the DS8000,
you must have your IBM Service Representative install Licensed Internal Code Version
2.4.3.65 or later on the Enterprise Storage Server. You also need DS8000 code bundle
6.0.500.52 or later on the DS8000.
4.3.1 Adding Enterprise Storage Server Copy Services Domain to the DS8000
Storage Complex
To migrate data using the DS Storage Manager from the Enterprise Storage Server to the
DS8000, add the Enterprise Storage Server Copy Services Domain server to the DS8000
Storage Complex. Adding the server can only be done using the DS8000 Storage Manager.
The DS8000 CLI does not require authentication like the DS Storage Manager does.
The DS8000 Storage Manager must authenticate to the Enterprise Storage Server before you
can issue any commands. The user ID and password used to log on to the DS Storage
Manager must be defined in the Enterprise Storage Server Specialist.

In Figure 4-5, the DS Storage Manager user ID is SLT-TEAM, which requires that the user ID
and password be created using the Enterprise Storage Server Specialist.
Figure 4-5 Enterprise Storage Server Specialist
To create the user ID and password, perform the following steps:

1. Click Enterprise Storage Server Specialist from the Enterprise Storage Server Launch
window. The Enterprise Storage Server Specialist window opens as shown in Figure 4-5.
2. Click Users. The User Administration window opens (Figure 4-6).
Figure 4-6 Enterprise Storage Server Specialist User Administration panel
3. Click Modify Users to add, remove, or modify existing user accounts.

4. Add the user ID and password that matches the DS Storage Manager user ID and
password as seen in Figure 4-7. Click Perform Configuration to create the user.
Figure 4-7 Enterprise Storage Server Specialist Modify Users panel
The new user is displayed in the User Administration window as shown in Figure 4-8.
Figure 4-8 Enterprise Storage Server Specialist User Administration panel with DS Storage Manager
ID
5. Connect to the Storage Manager and click Real-time manager  Manage Hardware 
Storage complexes.

6. In the Storage Complex window, click Select Action  Add 2105 Copy Services
Domain. A window prompting for the 2105 CS Domain Server 1 IP Address (Server A) is
displayed (Figure 4-9).
Figure 4-9 Adding a 2105 Copy Services Domain to the DS8000 Storage Complex
7. Enter the Server 1 address. You can enter a Server 2 IP addresses by checking the
Define a second Copy Services Server check box. Click OK.
8. Verify that the CS Domain has been added to the Storage Complex by viewing the Storage
Complexes after the task completes (Figure 4-10).
Figure 4-10 2105 Copy Services Domain added to DS8000 Storage Complex
The remaining steps for setting up Metro Mirror paths can be done with both the Storage
Manager and the DS8000 CLI. Both methods are shown in 4.3.2, “Creating Remote Mirror
and Copy paths” on page 54.
4.3.2 Creating Remote Mirror and Copy paths

Creating paths for Metro Mirror or Global Copy for data migration requires physical
connections between the Enterprise Storage Server and the DS8000. These paths can be
through a switch (including hardware for extended distances), or through direct physical
connections between the storage units.

In Metro Mirror, the Fibre Channel links can be connected by up to two switches. If the paths
are created through a switch, the ports for the storage units must be in an active zone set.
The zone configuration can be completed in several ways. In this example, two ports from
each storage unit are put in a single zone (Brocade) and activated as seen in Figure 4-11.
Figure 4-11 Enterprise Storage Server and DS8000 zone members
Remember: If channel extension technology is used for Metro Mirror links, make sure the
product used is supported in the environment (direct connect or SAN switch). Also ensure
that the SAN switch used is supported by the product vendor.
You need to know what the physical-to-logical layout of the I/O ports on both storage units is
so that you can set up the paths correctly. The chart in Figure 4-12 shows the numbering
scheme for the ports on the Enterprise Storage Server. There are four host bays, each with
four slots for adapters. The Enterprise Storage Server can have up to 16 host adapters,
allowing for a maximum of 16 Fibre Channel ports per Enterprise Storage Server.
Figure 4-12 Enterprise Storage Server I/O enclosures
The chart in Figure 4-13 on page 56 displays the scheme for the DS8000 in the first frame.
There are four I/O enclosures, each with six slots for adapters. Two slots in each enclosure
are reserved for the device adapters connected to the disk drive enclosures. That leaves four
slots for host adapters.

These adapters can either be ESCON or Fibre Channel. The Fibre Channel adapters can be
either 2 Gbps or 4 Gbps, and have four ports each. The slot numbers as shown Figure 4-13
on page 56 are logical slot numbers as seen in either DS Storage Manager or DS8000 CLI.
The logical number is one less than the physical number. The four ports on the Fibre Channel
adapters are labeled 0-3, and the numbering starts at the top port on each adapter.
Figure 4-13 DS8000 first expansion frame I/O enclosures

The DS8000 second expansion frame I/O enclosures are shown in Figure 4-14.
Figure 4-14 DS8000 second expansion frame I/O enclosures
Creating Remote Mirror and Copy paths with DS Storage Manager

After the physical paths are set up, the Metro Mirror/Global Copy paths can be created. To
create paths using DS Storage Manager, perform the following steps:
1. From the DS Storage Manager, click Real Time Manager  Copy Services  Paths.
2. In the Paths: Real-time window shown in Figure 4-15, select the Enterprise Storage
Server you added in 4.3.1, “Adding Enterprise Storage Server Copy Services Domain to
the DS8000 Storage Complex” on page 51 from the Storage complex list.
Figure 4-15 Paths window
3. Select the Source LSS from the Storage Unit list.

4. Click the Select Action box and select Create. The paths are being set up from the
Enterprise Storage Server to the DS8000 for the data migration. In this example, four
paths exist for LSS 17 (this area is empty if paths do not exist).
Tip: Selecting the source LSS is optional because the next window requires the LSS
selection. If it is selected, any existing paths are displayed.
5. In the Select source LSS window shown in Figure 4-16, select the source LSS you want
and click Next.
Remember: Each LSS to be used in the data migration requires paths to be created
one at a time using DS Storage Manager or the DS8000 CLI. A source LSS can have
multiple target LSSs, but the paths must be created for each target LSS separately.
In this example, LSS 17 is selected as the source LSS. Only the LSSs on the Enterprise
Storage Server are displayed in this list.
Figure 4-16 Create Paths - Source LSS

6. In the Select target LSS window, select the target LSS and click Next. Select the target
Storage Complex, Storage unit, Storage image, and LSS from the lists. Then select the
storage unit to generate the list of compatible LSSs to be displayed for selection.
In this example, LSS 10 is selected for this storage unit (Figure 4-17).
Figure 4-17 Select Target LSS for the DS8000
7. In the Select source I/O ports window as shown in Figure 4-18, select one or more source
ports and click Next.
The ports listed are available in the active zone set in the switch, or on the paths physically
connected between the two storage units. Figure 4-18 on page 59 shows that the ports
listed for the Enterprise Storage Server (Source) are the same as the zone members
shown in Figure 4-11 on page 55.
The ports are displayed by location rather than WWPN.
Figure 4-18 Select PPRC ports on Enterprise Storage Server
8. In the Select target I/O ports window as shown in Figure 4-19, select at least one target
port for each source port and click Next. As with the source ports, the target ports
available depends on how the zoning or cabling is set up. To select multiple target ports for
a single source port, press the Shift key while selecting the ports.

Tip: Have more than one path available for bandwidth and redundancy reasons. The
workload is balanced across the available paths by the code. Isolating the paths from
host I/O paths guarantees no interference from host I/O.
For Metro Mirror, use multiple paths due to timing sensitivity issues. Global Copy does
not have this sensitivity to shared host I/O ports and paths.
Each Source I/O port has a path available to both Target I/O ports due to the way the zone
was established on the switch. Both ports from both storage units are in a single zone.
With this selection, there are four logical paths for LSS 17 on the Enterprise Storage
Server, to LSS 10 on the DS8000.
Figure 4-19 Select PPRC ports on the DS8000
9. In the Select path options window (Figure 4-20 on page 61), select Define as
consistency group and click Next.
For Remote Mirror and Copy pairs, selecting the consistency group option supports the
consistent data between two LSSs (not a group of LSSs). Data consistency means that
the sequence of dependent writes is always kept in the copied data.
Tip: The consistency group option is not required for Global Copy paths in a data
migration scenario.
The Define as consistency group option itself can keep consistent data at the remote
site. In a rolling disaster, all volumes go into the queue full condition within the time interval
specified in the Consistency Group time-out value. The default time-out value is 120
seconds.
However, if all the volumes do not go into the queue full condition, use the commands
freezepprc and unfreezepprc to hold the I/O activity to the volumes not in the queue full
condition. You can also resume or release the held I/O without waiting for the Consistency
Group timeout to minimize the impact on the applications. These commands are issued at
the LSS level through the DS8000 CLI.

Remember: The freezepprc command removes the Remote Mirror and Copy paths
between the source LSS and target LSS.
For more information about using consistency groups, see IBM System Storage DS8000:
Copy Services in Open Environments, SG24-6788, and DS8000 Copy Services for IBM
System z, SG24-6787.
Figure 4-20 Consistency groups defined during PPRC paths creation
10.In the Verification window shown in Figure 4-21, review your selections carefully and then
click Finish. If changes need to be made, click Back to make the modifications before
clicking Finish.
Figure 4-21 Confirm the selections

11.After the selections are correct and the command completes, the created paths are
displayed as shown in Figure 4-22. Paths can be removed or added if necessary.
Figure 4-22 PPRC paths are displayed at completion
Creating Remote Mirror and Copy paths with DS8000 CLI

The Remote Mirror and Copy paths can also be created by connecting to the Enterprise
Storage Server using the DS8000 CLI. Log on to the Enterprise Storage Server using an
existing user ID and password on the Enterprise Storage Server, or using the one created for
DS Storage Manager.
Start the DS8000 CLI using the IP address of one cluster on the Enterprise Storage Server
and enter the user ID and password as shown in Example 4-1. A list of the storage units in the
Copy Services Domain Server is displayed. In this example, there are two storage units.
Example 4-1 Connect to Enterprise Storage Server using DS8000 CLI

C:\Program Files\ibm\dscli>dscli -hmc1 9.155.51.201
Enter your username: SLE-TEAM
Enter your password:
Date/Time: July 12, 2011 11:34:04 AM CET IBM DSCLI Version: 5.2.410.182 DS:
IBM.2105-22665
IBM.2105-22673
To create Remote Mirror and Copy paths, perform the following steps:
1. Before the paths can be created, you need to determine the remote WWNN and the
available paths. Use the lssi command to determine the remote WWNN as shown in
Example 4-2.
Example 4-2 Using the lssi command to obtain the WWNN of the remote system
dscli> lssi
Date/Time: July 12, 2011 11:46:06 AM CET IBM DSCLI Version: 5.2.410.182
Name ID Storage Unit Model WWNN State ESSNet
==================================================================================
DS8k-SLE05 IBM.2107-75L4741 IBM.2107-75L4740 931 5005076305FFC786 Online Enabled

2. Display the available PPRC ports between the two storage units using the lavailpprcport
command as seen in Example 4-3. The -remotewwnn parameter is the WWNN determined
in the previous step. Use the Enterprise Storage Server (2105 in this example) as the -dev
parameter, and the DS8000 (2107 in this example) as the -remotedev parameter. This
information matches what was seen for the DS Storage Manager. They are identical
because of the active zone created at the beginning of this section and shown in
Figure 4-11 on page 55.
Example 4-3 Query available PPRC ports between Enterprise Storage Server and DS8000
dscli> lsavailpprcport -dev IBM.2105-22673 -remotedev IBM.2107-75L4741
-remotewwnn 5005076305FFC786 -fullid 17:10
IBM.2105-22673
Local Port Attached Port Type
================================================
IBM.2105-22673/I0004 IBM.2107-75L4741/I0140 FCP
IBM.2105-22673/I0004 IBM.2107-75L4741/I0142 FCP
IBM.2105-22673/I00AC IBM.2107-75L4741/I0140 FCP
IBM.2105-22673/I00AC IBM.2107-75L4741/I0142 FCP
3. Use the WWNN and a port pair to create a path using the mkpprcpath command from the
Enterprise Storage Server (Example 4-4). Verify that the information entered is correct
before running the command. In this example, four paths are created between LSS 17 on
the Enterprise Storage Server and LSS 10 on the DS8000.
Example 4-4 Creating PPRC paths between Enterprise Storage Server and DS8000
dscli> mkpprcpath -dev IBM.2105-22673 -remotedev IBM.2107-75L4741 -remotewwnn
5005076305FFC786 -srclss 17 -tgtlss 10 I0004:I0140 I00AC:I0140 I0004:I0142
I00AC:I0142
IBM.2105-22673
CMUC00149I mkpprcpath: Remote Mirror and Copy path 17:10 successfully
established.
4. Query and verify the paths using the lspprcpath command as shown in Example 4-5. If
you need to make changes, remove the path and recreate or modify it using DS Storage
Manager. The output lists the Enterprise Storage Server information in the Src, SS, and
Port columns, and the DS8000 information in the Tgt column.
Example 4-5 Query PPRC paths between Enterprise Storage Server and DS8000
dscli> lspprcpath -dev IBM.2105-22673 17
IBM.2105-22673
Src Tgt State SS Port Attached Port Tgt WWNN
=========================================================
17 10 Success FF10 I00AC I0140 5005076305FFC786
17 10 Success FF10 I00AC I0142 5005076305FFC786
17 10 Success FF10 I0004 I0140 5005076305FFC786
17 10 Success FF10 I0004 I0142 5005076305FFC786

Creating paths with TSO
To establish a path between the Enterprise Storage Server and the DS8000 for Metro Mirror,
use the CESTPATH command. You must know the SSID, WWNN, and LSS number for the
primary and remote storage units. These numbers can be displayed with the CQUERY
command.
The example shown in Example 4-6 establishes a path between the Enterprise Storage
Server and the DS8000. The SSID of the primary volume is x’1710’, the WWNN is
5005076300C09629, and the LSS is x’00’. The SSID of the secondary volume is x’1711’, the
WWNN is 5005076305FFC786, and the LSS is x’01’. In this example, the consistency group
option is set to NO.
Example 4-6 TSO CESTPATH output

CESTPATH DEVN(X'2028') -
PRIM(X'1710' 5005076300C09629 X'00') -
SEC(X'1711' 5005076305FFC786 X'01') -
LINK(X'00300200') CGROUP(NO)
Creating paths with ICKDSF

The PPRCOPY ESTPATH command is used to establish Remote Mirror and Copy paths
between the primary and remote LSSs. Each command can establish up to eight paths.
In Example 4-7, the FCPP parameter specifies up to eight paths. Each path is an 8-digit
hexadecimal address in the form x’aaaabbbG67 b’. In this form, aaaa is the primary system
adapter (SAID) and bbbb is the remote system adapter ID (SAID). The World Wide Node
Name (WWNN) for the primary and remote are specified in the WWNN parameter, with the
primary listed first followed by the remote.
Example 4-7 ICKDSF ESTPATH output

//IKJEFT01 JOB MSGLEVEL=(1,1),MSGCLASS=A,NOTIFY=user id
//STEP01 EXEC PGM=ICKDSF
//SYSPRINT DD SYSOUT=*
//VOL1 DD UNIT=3390,VOL=SER=DS6400,DISP=SHR
//SYSIN DD *
PPRCOPY ESTPATH DDNAME(VOL1) -
FCPP(X'00000100') PRI(X'0002',AAVCA) -
SEC(X'0003',AAVCA) WWNN(005076300C09629,5005076305FFC786) -
CGROUP(NO) LSS(X'00',X'01')
4.3.3 Creating Remote Mirror and Copy pairs

When you are creating pairs for Metro Mirror or Global Copy, you must take into account the
size of the volumes to be copied from the Enterprise Storage Server. The DS8000 default
volume is created as 2^30 bytes, but the Enterprise Storage Server volumes are created as
10^9 bytes.
When creating fixed block volumes on the DS8000, you have three size choices:
򐂰 ds: The number of bytes allocated will be the requested capacity value times 230 .
򐂰 Enterprise Storage Server: The number of bytes allocated will be the requested capacity
value times 109 .
򐂰 blocks: The number of bytes allocated will be the requested capacity value times 512
bytes (each block is 512 bytes).

To create the pairs, the volume on the DS8000 must be as large or larger than the volume on
the Enterprise Storage Server. If the volumes on the DS8000 are larger than the Enterprise
Storage Server, the extra space is not used. However, if you might use the Enterprise Storage
Server as a remote system after the migration, make the volumes the same size. For fixed
block volumes, create the DS8000 volumes using the Enterprise Storage Server type.
For more information, see IBM System Storage DS8000: Copy Services in Open
Environments, SG24-6788.
CKD volumes have the same considerations about volume size. CKD volumes are specified
in number of cylinders. The volumes on the DS8000 must have the same or greater number
of cylinders as the volumes on the Enterprise Storage Server.
For more information, see DS8000 Copy Services for IBM System z, SG24-6787.
Another important aspect to consider before creating the configuration for the data migration
to the DS8000 is the volume address differences between the two storage units. On the
Enterprise Storage Server, open systems volume IDs are given in an 8-digit format,
xxx-sssss. In this form, xxx is the LUN ID and sssss is the serial number of the Enterprise
Storage Server. When referring to these volumes with DS8000 CLI, add 1000 to the volume
ID. Remember the limitations on the Enterprise Storage Server address ranges shown in
Figure 4-23.
0000 to 0FFF CKD volumes (4096 possible addresses)

1000 to 1FFF Open systems fixed block (FB) LUNs (4096 possible addresses)
Figure 4-23 Enterprise Storage Server LCU/LSS restriction
On the DS8000, the range of available addressing is significantly greater than on the
Enterprise Storage Server. The entire storage unit can be configured for just CKD or just FB
using LUNs 00-FF on LSSs/LCUs 00-FE (FF is reserved).
If the configuration on the DS8000 is a mixed CKD and FB environment, the CKD or FB
volumes must be contained within a grouping of 16 LSSs/LCUs. For example, CKD volumes
are configured in 2000-2FFF and FB volumes in 3000-3FFF, where 20-2F and 30-3F are the
range of 16 LCUs/LSSs.
No restrictions exist for the creation of the LSS/LCU ranges other than to group by 16. The 16
groups can be all CKD, all FB, or mixed, as long as any single group is the same type. Take
grouping into account when planning the DS8000 configuration because the storage unit
enforces the groupings.
After you complete the volume configuration on the DS8000 compatible with the Enterprise
Storage Server, the volumes will be formatted by an internal DS8000 process. Otherwise the
volumes cannot be used as target volumes on the DS8000. This formatting must be complete
before creating the pairs. The time needed for the volume initialization completion varies
depending on the size.
Remember: If you attempt to use the volumes before the volume initialization has
completed, the establish of the copy pairs fails. This is an expected result in this case.

Creating Remote Mirror and Copy pairs with DS Storage Manager
To create the Remote Mirror and Copy pairs using the DS Storage Manager, perform the
following steps:
1. Open the Metro Mirror wizard using PROCEDURE.
2. In the initial window of the wizard (Figure 4-24), select the Storage Complex, the Storage
Unit, and the Resource Type (LSS or Show All Volumes).
Figure 4-24 Select the Enterprise Storage Server and LSS
3. If you selected LSS in the Resource Type list, select the specific LSS in the Specify LSS
list. If you selected Show All Volumes, select All FB Volumes or All CKD Volumes.
4. Select the volumes to be used, and click the Select Action list and select Create.
In this example, the selection is being made by LSS. Specify LSS 17 (which we have
already made paths for) volume 1700.
5. In the Volume Pairing Method window, select the method for Volume Pairing:
– Automated volume pair assignment automatically pairs the first selected source volume
with a target volume of the same size. All subsequent pairs are automatically assigned
based on compatible size in a sequential fashion. The lowest source volume number
are paired with the lowest target volume number.

– Manual volume pair assignment requires a manual selection of all source and target
volumes. This method is used in the example (Figure 4-25). Repeat the process for
each selected source volume.
Figure 4-25 Volume Pairing Method selection
6. The Select source volumes window displays the available source volumes based on an
LSS. You can also create paths from this panel by clicking Create Paths, which starts the
Paths wizard.
In this example, the source volumes are in LSS 17 on the Enterprise Storage Server.
Select the source volume you want and click Next (Figure 4-26).
Figure 4-26 Select the Enterprise Storage Server source volumes

7. In the Source ID window, select the target Storage complex, Storage unit, Storage image,
and Resource type as shown in Figure 4-27. The example uses the DS8000 LSS 10
volume 1010 as the target device.
Figure 4-27 Select DS8000 target volumes
8. Select the volume for the pair and click OK.

9. In the Select copy options window, select the option for creating the pairs as Metro Mirror
or Global Copy (Figure 4-28).
Figure 4-28 Select Metro Mirror or Global Copy options

Additional options are available for selection in this panel:
– Reset reservation (Open Systems)
– Perform initial copy
– Permit read access from target
The remaining three options have a dependency on the type of copy options selected.
Perform initial copy allows the selection of Suspend Metro Mirror relationships after
initial copy. The remaining two options are available only for CKD pairs.
Click Next after making your selections.
10.The Verification window prompts for confirmation of the selected configuration of pairs. If
you need to make any changes, click Back and make them before returning to this
window. When ready, click Finish to complete the pair selection as seen in Figure 4-29.
Figure 4-29 Confirm the selection
11.After the pairs have been created, verify the state of the relationship by checking the
Metro Mirror: Real-time window for the defined source volumes (Figure 4-30).
Figure 4-30 Check state of the pairs
Remote Mirror and Copy pairs are created one LSS at a time. All the volumes in an LSS can
be used, but the process must be repeated for each LSS involved in the migration.

Creating Remote Mirror and Copy pairs with DS8000 CLI
After the Remote Mirror and Copy paths are created, create Metro Mirror and Global Copy
pairs between the Enterprise Storage Server and the DS8000. Use the mkpprc command to
create pairs as shown in Example 4-8. In this example, a single Metro Mirror pair is created
between source volume 1700 on the Enterprise Storage Server and target volume 1010 on
the DS8000.
Example 4-8 mkpprc with DS8000 CLI

dscli> mkpprc -dev IBM.2105-22673 -remotedev IBM.2107-75L4741 -type mmir 1700:1010
IBM.2105-22673
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 1700:1010
successfully created.
Requirement: The DS8000 CLI must be used from the Enterprise Storage Server to
create the Remote Mirror and Copy pairs.
Example 4-9 illustrates the creation of Global Copy pairs with DS8000 CLI on the same
volumes used in the preceding examples. Before starting this example, the Metro Mirror pair
was removed.
Example 4-9 Create Global Copy pairs

dscli> mkpprc -dev IBM.2105-22673 -remotedev IBM.2107-75L4741 -type gcp 1700:1010
IBM.2105-22673
Creating the Remote Mirror and Copy pairs can be tedious task if many volumes are involved
in the migration. Up to 4096 source volumes are possible on the Enterprise Storage Server.
An easy way to create many pairs is to use the DS8000 CLI in a scripting mode. A script to
create Global Copy pairs is shown in Figure 4-31.
#> cat mkpprc_script

# PPRC for IBM1->IBM2 CKD pairs
dscli -cfg IBM1_IBM2 -user admin -passwd xxxxx mkpprc -dev IBM.2105-22673
-remotedev IBM.2107-1303151 -type gcp 0000-003f:c800-c83f 0100-013f:c900-c93f
0200-023f:ca00-ca3f 0300-033f:cb00-cb3f 0400-043f:cc00-cc3f
0500-053f:cd00-cd3f 0600-063f:ce00-ce3f 0700-073f:cf00-cf3f
0800-083f:e800-e83f 0900-093f:e900-e93f 0a00-0a3f:ea00-ea3f
0b00-0b3f:eb00-eb3f 0c00-0c3f:ec00-ec3f 0d00-0d3f:ed00-ed3f
0e00-0e3f:ee00-ee3f 0f00-0f3f:ef00-ef3f
#>
Figure 4-31 DS8000 CLI script example for mkpprc
This example shows creating 1024 Global Copy pairs. The source volumes are on the
Enterprise Storage Server (which has 64 volumes in each LCU) and pairs are created with
corresponding target volumes on the DS8000. The pairs are created as a single pair of LSSs
at a time. For example, source LSS 00 with volumes 00-3F is paired with target LSS C8 with
volumes 00-3F as 0000-003F:C800-C83F.

Tip: Using the scripting method is an ideal way to verify that pairs are set up correctly
before the pairs are created. It is also convenient for creating many pairs.
Creating pairs with TSO

The CESTPAIR command is used to establish Metro Mirror and Global Copy relationships
between the Enterprise Storage Server and the DS8000. This command is used to indicate
what kind of an operation to perform:
򐂰 An initial establish of volumes that were in the simplex state
򐂰 A resynchronization of a suspended pair of volumes
򐂰 A Failover/Failback process
In this example, you are establishing an initial copy from a simplex state.
The option parameter OPTION has two mutually exclusive values: SYNC and XD. SYNC is
specified to create Metro Mirror pairs. XD is used to specify Global Copy pairs.
In Example 4-10, the primary volume is in LSS x’00’ on the Enterprise Storage Server. The
primary volume has the following characteristics:
򐂰 The SSID of the LSS is x’1710’
򐂰 The serial number of the Enterprise Storage Server is 22673
򐂰 The CCA is x’28’
򐂰 The LSS is x’00’
The remote volume is in LSS x’01’ on the DS8000, and has the following characteristics:
򐂰 The SSID of the LSS is x’1711’
򐂰 The serial number of the DS8000 is L4741
򐂰 The CCA is x’28’
򐂰 The LSS is x’01’.
The MSGREQ(YES) parameter specifies that Metro Mirror waits until the initial full volume
copy operation is complete before issuing the completion message.
Example 4-10 TSO CESTPAIR Metro Mirror

CESTPAIR DEVN(X'2028') -
PRIM(X'1710' 22673 X'28' X'00') -
SEC(X'1711' L4741 X'28' X'01') -
OPTION(SYNC) MODE(COPY) ONLINSEC(NO) MSGREQ(YES)
In order for a pair to be created as Global Copy, specify whether the pair comes from the
simplex or suspended state. This means an initial copy of a newly established pair (simplex)
or a resynchronization of a suspended pair. The MODE parameter is used to specify either
COPY or RESYNC. The CESTPAIR command in Example 4-11 includes the OPTION(XD)
and MODE(COPY) to signify Global Copy from a simplex state.
Example 4-11 TSO CESTPAIR Global Copy

CESTPAIR DEVN(X'400A') OPTION(XD) MODE(COPY) -
PRIM(X'4000' ABTV1 X'0A' X'00') -
SEC(X'8000' 20781 X'8A' X'80')

Creating pairs with ICKDSF
The ESTPAIR command is used to establish Metro Mirror and Global Copy relationships
between the Enterprise Storage Server and the DS8000 volumes. In Example 4-12, the pair
is created for Metro Mirror (the OPTION(SYNC) parameter). To create the pair as Global
Copy, use OPTION(XD).
Example 4-12 ISCKDSF ESTPAIR

//SYSIN DD *
PPRCOPY ESTPAIR DDNAME(VOL1) PRI(X'0002',AAVCA,X'00') -
SEC(X'0003',22673,X'00') MODE(COPY) MSGREQ(NO) OPTION(SYNC) -
CRIT(NO) LSS(X'00',X'01')
4.3.4 Completing the data migration

This section includes the steps taken with DS8000 CLI, TSO, ICKDSF, and DS Storage
Manager to move the production site on to the new storage. Moving the host systems is
covered in 4.5, “Post-migration tasks” on page 90.
After the Remote Mirror and Copy pairs are established, the data will start copying. For Metro
Mirror, wait for the pairs to enter a Full Duplex state so that the data migration can be
completed. For Global Copy, there is an intermediate state required to get to the Full Duplex
state.
The following steps synchronize the data from the Enterprise Storage Server to the DS8000.
These steps are common for any interface used to manage the Remote Mirror and Copy
relationships.
1. Verify full duplex mode (Metro Mirror) or out-of-sync tracks (Global Copy) are at or near
zero.
2. If you are using Global Copy, convert to synchronous (go-to-sync function).
3. Suspend the source I/O.
4. Suspend pairs.
5. Delete pairs.
Completing the data migration with the DS8000 CLI

To complete the migration with the DS8000 CLI, perform the following steps:
1. Use the DS8000 CLI command lspprc to query the state of the Metro Mirror pairs as
shown in Example 4-13. Only one pair is shown in this example. The current state is Copy
Pending, which means the copy is in progress. The copy is complete when the state
changes to Full Duplex.
Example 4-13 Monitor state of Metro Mirror pairs

dscli> lspprc -dev IBM.2105-22673 1700
IBM.2105-22673
ID State Reason Type SourceLSS Timeout (secs) Critical Mode
First Pass Status
==================================================================================

1700:1010 Copy Pending - Metro Mirror 17 120 Disabled
Invalid
2. The same command is used to query the Global Copy relationship. Query the pairs to
monitor the Out of Sync Tracks, as seen in Example 4-14. In this example, the local and
remote copies still have many tracks to copy (288991) and the First Pass has not yet
completed.
Example 4-14 Non-zero Out of Sync Tracks

dscli> lspprc -l -dev IBM.2105-22673 1700
Date/Time: July 14, 2011 4:07:53 PM CET IBM DSCLI Version: 5
ID State Reason Type Out Of Sync Tracks ... First Pass Status
==================================================================================
1700:1010 Copy Pending - Global Copy 288991 ... ... ... ... False
As this number approaches zero (or gets to zero) and the First Pass field becomes true,
the go to sync function is started. This state must be reached for each pair in the data
migration. Example 4-15 shows the Out of Sync Tracks at zero (0) and the First Pass
Status at true.
Example 4-15 Out of Sync Tracks at zero

Date/Time: July 14, 2011 4:17:40 PM CET IBM DSCLI Version:
=================================================================================
1700:1010 Copy Pending - Global Copy 0 .... .... .... ..... True
3. This Global Copy pair is now ready for the final copy to be performed. Before running this
command, stop the host I/O and synchronize all the data to disk to avoid losing any
updates after the relationship is removed.
4. Use the mkpprc command to start the go to sync function on the same pair, but change
the -type option to mmir (Example 4-16).
Example 4-16 Go to Sync function

dscli> mkpprc -dev IBM.2105-22673 -remotedev IBM.2107-75L4741 -type mmir 1700:1010
Date/Time: July 14, 2011 4:18:16 PM CET IBM DSCLI Version: 5.2.410.182 DS:
IBM.2105-22673
5. Query the pair using the lspprc command to verify that the pair has completed the copy
when the state changes to Full Duplex (Example 4-17).
Example 4-17 Full Duplex state

Date/Time: July 14, 2011 4:18:35 PM CET IBM DSCLI Version:
ID State Reason Type Out Of Sync Tracks
============================================================
1700:1010 Full Duplex - Metro Mirror 0

6. The copy is now complete and the I/O is all stopped. Remove the relationship with the
rmpprc command as shown in Example 4-18.
Tip: Using this command with multiple pairs in multiple LSSs will result in a
confirmation question for each range of pairs. The -quiet option can be included to turn
off this confirmation prompt.
Example 4-18 Remove the relationship

dscli> rmpprc -dev IBM.2105-22673 -remotedev IBM.2107-75L4741 1700:1010
IBM.2105-22673
CMUC00160W rmpprc: Are you sure you want to delete the Remote Mirror and Copy
volume pair relationship 1700:1010:? [y/n]:y
CMUC00155I rmpprc: Remote Mirror and Copy volume pair 1700:1010 relationship
successfully withdrawn.
7. You are asked to confirm the removal of the pair. Respond y (for yes) to delete the
relationship, or n (for no) to cancel.
Complete these steps for all pairs involved in the data migration.
Completing the migration with TSO

To complete the migration with TSO, perform the following steps:
1. The CQUERY command is used to query the status of one volume of a pair. It can also be
used to query all paths that are associated with the LSS for the specified device number.
This command can be issued to either a primary or remote volume. A host system
attached to only the primary volume cannot obtain the status of the secondary volume for
that pair.
The output listed in Example 4-19 is from the CQUERY command issued to volume
x’400A’. The command was issued specifically for a Global Copy pair.
Notice that the STATE is PENDING.XD to signify Global Copy. In addition, the Out of Sync
Tracks are listed in this example as 119475. The total number of tracks on the volume and
the Percent Copy Complete are listed below.
This information is not listed for volumes with all tracks in sync. The path information is
shown only when the CQUERY command is issued to a primary volume. In this example,
two paths are viable.
Although the example is for a Global Copy relationship, a Metro Mirror pair displays
equivalent information during the copy phase. However, the STATE is displayed as
PENDING instead of PENDING.XD.
Example 4-19 TSO CQUERY command output

ANTP8802I CQUERY DEVN(X'400A')
ANTP0090I CQUERY FORMATTED LVL 3 881
VOLUME REPORT
************** PPRC REMOTE COPY CQUERY - VOLUME ********************
* (PRIMARY) (SECONDARY) *
* SSID CCA LSS SSID CCA LSS*
*DEVICE LEVEL STATE PATH STATUS SERIAL# SERIAL# *
*------ --------- ---------- ----------- --------- --------- *
* 400A PRIMARY.. PENDING.XD ACTIVE.. 4000 0A 00 8000 8A 80 *
* CRIT(NO)....... CGRPLB(NO). 0000000ABTV1 000000020781*

* PATHS PFCA SFCA STATUS: DESCRIPTION *
* ----- --------- ------ ------------------- *
* 2 0010 0143 13 PATH ESTABLISHED... *
* 0031 0102 13 PATH ESTABLISHED... *
* ---- ---- 00 NO PATH............ *
* ---- ---- 00 NO PATH............ *
* IF STATE = PENDING/SUSPEND: TRACKS OUT OF SYNC = 119475 *
* TRACKS ON VOLUME = 150255 *
* PERCENT OF COPY COMPLETE = 21% *
* SUBSYSTEM WWNN LIC LEVEL *
* ----------- ---------------- ----------- *
* PRIMARY.... 5005076300C09629 2.4.40.0045 *
* SECONDARY.1 5005076305FFC786 *
********************************************************************
ANTP0001I CQUERY COMMAND COMPLETED FOR DEVICE 400B. COMPLETION CODE: 00
Tip: The CESTPAIR command does not support the go-to-sync and suspend operation
(ICKDSF, DS8000 CLI, and DS Storage Manager do). When using the TSO interface,
you can set the process to trigger when the system issues a state change message that
the duplex state is reached.
2. Query the status of the target volumes using the CQUERY command as shown in
Example 4-20. Notice that the state of the pair is PENDING.XD and no path information is
displayed. This is a normal response for the target storage unit.
Example 4-20 Query target status

ANTP8802I CQUERY DEVN(X'808A')
ANTP0090I CQUERY FORMATTED LVL 3 943
VOLUME REPORT
*DEVICE LEVEL STATE PATH STATUS SERIALÄ SERIALÄ *
*------ --------- ---------- ----------- --------- --------- *
* 808A SECONDARY PENDING.XD ACTIVE.. 4000 0A 00 8000 8A 80 *
* ............... ........... 0000000ABTV1 000000020781*
* PATHS SAID DEST STATUS: DESCRIPTION *
* ----- --------- ------ ------------------- *
* 0 ---- ---- 00 NO PATH............ *
* ---- ---- 00 NO PATH............ *
* ---- ---- 00 NO PATH............ *
* ---- ---- 00 NO PATH............ *
* ----------- ---------------- ----------- *
* PRIMARY.... 5005076300C09629 5.2.400.0064 *
* SECONDARY.1 5005076305FFC786 *
********************************************************************
3. Before removing relationships, stop all host activity and ensure that all data is written to
disk. These steps ensure that a full and complete copy exists on the DS8000. After that,
move the pairs to a synchronous state, complete the copy, and attach the application to
the target storage unit.

4. The pair is ready for synchronization after all the tracks have been copied. Use the
CESTPAIR command with the SYNC option on the same pair (Example 4-21).
Example 4-21 CESTPAIR to synchronize the pair

CESTPAIR DEVN(X'400A') OPTION(SYNC) MODE(RESYNC) -
PRIM(X'4000' ABTV1 X'0A' X'00') -
SEC(X'8000' 20781 X'8A' X'80')
Monitor the copy, using the CQUERY command until the pair reaches a Duplex state. When
this state is reached, the copy is complete.
5. Use the CSUSPEND command to suspend and remove the pairs as shown in Example 4-22.
Query the state to confirm that the pairs are suspended using the CQUERY command.
The state will be SUSPEND(3), which means the Global Copy was suspended by a host
command to the source storage unit.
Example 4-22 CQUERY to suspend the pair

CSUSPEND DEVN(X'400A') -
PRIM(X'4000' ABTV1 X'0A' X'00') -
SEC(X'8000' 20781 X'8A' X'80')
6. The copy is now complete and the pairs are suspended. Delete the pairs by issuing the
CDELPAIR command to the source storage volumes (Example 4-23).
Example 4-23 CDELPAIR to delete the pair

CDELPAIR DEVN(X'400A') -
PRIM(X'4000' ABTV1 X'0A' X'00') -
SEC(X'8000' 20781 X'8A' X'80')
After these steps are performed for all pairs involved in the data migration, you are ready to
move the host and application to the target DS8000.
Completing the migration with ICKDSF

To complete the migration with ICKDSF, perform the following steps:
1. Use the PPRCOPY QUERY command to check the state of pairs as shown in
Example 4-24.
Example 4-24 ICKDSF query pair

//STEP01 EXEC PGM=ICKDSF,TIME=30 ,PARM='NOREPLYU'
//SYSIN DD *
PPRCOPY QUERY DDNAME(Vol1)
PPRCOPY QUERY DDNAME(Vol1) PATHS
When the pair reaches either of the following states, you are ready to suspend the host
application and flush all writes to disk:
If the pair created is a Metro Mirror relationship and reaches DUPLEX state, or it is a Global
Copy relationship and the out-of-sync tracks number is near zero,
2. A Global Copy relationship must be converted to synchronous to complete the copy. Use
the PPRCOPY CESTPAIR command, changing the parameter OPTION(XD) to

OPTION(SYNC) and adding the parameter MODE(RESYNC). The relationship is shown
in Example 4-25.
Example 4-25 ICKDSF go-to-sync

PPRCOPY ESTPAIR UNIT(4080) LSS(X’01’,x’03’) -
PRI(X’A001’,22673,X’00’) SEC(X’2801’,L4741,X’00’) -
OPTION(SYNC) MODE(RESYNC)
3. Use the PPRCOPY SUSPEND command to suspend the Metro Mirror relationship before you
remove it as shown in Example 4-26. Removing the mirror ensures all the data has been
copied from the source to the target storage unit.
Example 4-26 ICKDSF suspend pair

//SYSIN DD *
PPRCOPY SUSPEND DDNAME(VOL1) PRI(X'0002',22673,X'00') -
SEC(X'0003',AAVCA,X'00') LSS(X'00',X'01')
4. Delete the pair using PPRCOPY DELPAIR command as shown in Example 4-27.
Example 4-27 ICKDSF delete pair

//SYSIN DD *
PPRCOPY DELPAIR DDNAME(VOL1) PRI(X'0002',22673,X'00') -
SEC(X'0003',AAVCA,X'00') LSS(X'00',X'01')
After performing these steps for all pairs involved in the data migration, you are ready to move
the host and application to the target DS8000.
Completing the migration with DS Storage Manager

To complete the data migration using DS Storage Manager, complete the following steps:
1. To monitor the out-of-sync tracks, click Real-time manager  Copy services  Metro
Mirror  Select storage specifics.
2. Select the pair to monitor and click the Out-of-sync tracks option.

Figure 4-32 shows an example of querying a pair for out-of-sync tracks. The out-of-sync
tracks are at zero (0). This means that you are ready to complete the data migration from
the storage units and move the applications to the new storage.
Figure 4-32 Zero out-of-sync tracks
The example in Figure 4-33 shows the out-of-sync tracks as non-zero. Monitor this state
until it gets to the state shown in Figure 4-32.
Figure 4-33 Non-zero out-of-sync tracks

3. For a Global Copy relationship, select the pairs that are now ready to convert to sync
(out-of-sync tracks are at zero or close to zero). Using the same path previously
mentioned, select the Convert to synchronous option in the Select Action list as shown
in Figure 4-34.
Figure 4-34 Convert to synchronous
4. In the next window, select which system to suspend the Global Copy relationship at and
select Suspend from the Select Action list (Figure 4-35).
Figure 4-35 Suspend Metro Mirror

5. In the Select source or target for volumes to be suspended window, select Suspend at
source as shown in Figure 4-36.
Figure 4-36 Suspend at source selected
6. Click OK to confirm the selection as shown in Figure 4-37.
Figure 4-37 Confirm the selection
7. After the command completes, query the state of the relationship. As seen in Figure 4-38,
the pairs are now in full duplex state. After stopping the host I/O and confirming that all the
out-of-sync tracks are at zero, you are ready to remove the relationship.
Figure 4-38 Metro Mirror Full Duplex
Important: Do not remove the relationship if the out-of-sync tracks are non-zero.
Removing the pairs any earlier would result in having incomplete data on the remote
system.

8. Select the pair involved in the migration, and select Suspend from the Select Action list.
9. Select the pair involved in the migration again, and select Delete the Select Action list as
shown in Figure 4-39.
Figure 4-39 Delete Metro Mirror pairs
A confirmation warning for the deletion displays (Figure 4-40).
Figure 4-40 Confirm Delete
The data migration using IBM Copy Services is complete. The remaining steps for bringing
the applications up on the new DS8000 are listed in 4.5, “Post-migration tasks” on page 90.
4.4 Data migration from DS8000 to DS8000

The steps needed for migrating data between two DS8000s are similar to those for migrating
data from Enterprise Storage Server to DS8000. The Copy Services steps can be started
from either the local or a remote storage unit.
If the DS Storage Manager is used for data migration, the storage unit of one system must be
added to the storage complex of the other. This process is described in 4.3.1, “Adding
Enterprise Storage Server Copy Services Domain to the DS8000 Storage Complex” on
page 51.

The steps needed for moving the application to the new storage are the same regardless of
what the source storage unit is. For more information, see 4.5, “Post-migration tasks” on
page 90.
Because the commands for Enterprise Storage Server and DS8000 as source storage are
similar, the migration between two DS8000s is covered using the DS8000 CLI only. The
following steps are described:
򐂰 Configuring the remote DS8000
򐂰 Creating Remote Mirror and Copy paths
򐂰 Creating Remote Mirror and Copy pairs
򐂰 Completing the data migration from DS8000 to DS8000
4.4.1 Configuring the remote DS8000

The remote storage unit can be configured by using DS Storage Manager (Real-time and
Simulated), or by using DS8000 CLI. The volumes on the remote system must match or be
larger than the volume sizes and types on the local system. As noted previously, if the target
volumes are created larger than the source volumes, the extra space is not used. In addition,
if the system that the data is being migrated from will ever be used as a Remote Mirror and
Copy remote for the new DS8000, the volumes must have matching sizes.
The examples in this section use DS8000 CLI for performing the configuration. To configure
the remote DS8000, follow these steps:
1. Confirm the logical configuration on the source storage unit using the lsfbvol command
as shown in Example 4-28. For simplicity reasons, the fb volumes in volume group V3 are
listed as the source volumes. Notice that the capacities of these volumes are 5 GB and 1
GB (DS sizes).
Example 4-28 Check the logical configuration on local

dscli> lsfbvol -volgrp v20 -dev IBM.2107-75ABTV1 4204-4206
Date/Time: July 8, 2011 1:16:40 PM PDT IBM DSCLI Version: 7.6.10.511 DS:
IBM.2107-75ABTV1
Name ID accstate datastate configstate deviceMTM datatype extpool cap
(2^30B)
=================================================================================
TEAMRED_A 4204 Online Normal Normal 2107-900 FB 512 P4 5.0
2. Issue the mkfbvol command to configure the target DS8000 to match the source DS800
as shown in Example 4-29. You need to issue the command once for each size of volume.
In this example, the first command creates two 5 GB volumes with the same nickname
(pprc_tgt_tic6) and assigns both to the volume group V20. The second command is the
same except for the size (1 GB).
Example 4-29 Create matching configuration on remote

dscli> mkfbvol -dev IBM.2107-7520781 -extpool p0 -cap 5 -type ds -name
pprc_tgt_tic6 -volgrp v20 E404-E405
IBM.2107-7520781
CMUC00025I mkfbvol: FB volume E404 successfully created.

dscli> mkfbvol -dev IBM.2107-7520781 -extpool p0 -cap 1 -type ds -name
pprc_tgt_tic6 -volgrp v20 E406
IBM.2107-7520781
Reminder: Volumes created with DS8000 CLI in a single command will be the same size
and from the same extent pool. Volumes with even LSSs are created in even extent pools.
Volumes with odd LSSs are created in odd extent pools.
3. Verify the configuration on the target DS8000 using the lsfbvol command
(Example 4-30).
Example 4-30 Verify the logical configuration on the remote

dscli> lsfbvol -dev IBM.2107-7520781 E404-E406
IBM.2107-7520781
Name ID accstate datastate configstate deviceMTM datatype extpool cap
(2^30B)
=================================================================================
pprc_tgt_tic6 E404 Online Normal Normal 2107-900 FB 512 P0 5.0
Repeat these steps until the configuration of the target DS8000 matches the configuration on
the volumes used in the data migration. Remember to plan the layout of the volumes based
on performance considerations and opportunities for growth.
4.4.2 Creating Remote Mirror and Copy paths

To create the Remote Mirror and Copy paths for the data migration using the DS8000 CLI,
perform the following steps:
1. Use the lssi command to discover the device ID and WWNN of the remote system as
shown in Example 4-31.
The information listed under ID is the device ID that will be used. In this example, the
device ID is IBM.2107-7520781. The model is 922. The WWNN is 5005076303FFC1A5.
Example 4-31 Query device ID and WWNN using lssi command

dscli> lssi
Date/Time: July 8, 2011 2:09:22 PM PDT IBM DSCLI Version: 7.6.10.511 DS: -
=================================================================================
- IBM.2107-7520781 IBM.2107-7520780 922 5005076303FFC1A5 Online Enabled
If the remote system is a logically partitioned (LPAR) storage unit, the same lssi
command can be used. However, it returns the information for both LPARs (also known as
Storage Facility Images or SFI). You must know which SFI the data will be migrated to.

The output for an LPAR system is shown in Example 4-32. Here the device IDs are
IBM.2107-75ABTV1 and IBM.2107-75ABTV2. The model is 9A2, with both WWNNs listed for
each SFI.
Example 4-32 Query device ID and WWNN on an LPAR storage unit

dscli> lssi
Date/Time: July 8, 2011 2:12:34 PM PDT IBM DSCLI Version: 7.6.10.511 DS: -
=================================================================================
- IBM.2107-75ABTV1 IBM.2107-75ABTV0 9A2 5005076303FFC663 Online Enabled
- IBM.2107-75ABTV2 IBM.2107-75ABTV0 9A2 5005076303FFCE63 Online Enabled
2. After the zoning and connectivity are ready, query which paths are available to use for
Remote Mirror and Copy using the lsavailpprcport command. Issue this command from
the local system to the remote.
In Example 4-33, the command queries the possible ports between the local system
IBM.2107-75ABTV1 and the remote system IBM.2107-7520781, between LSS 42 and E4. As
you can see in the output, there are two available paths configured. Notice that each
attached port (remote) is visible to each local port.
Example 4-33 Check for available paths between the storage units
dscli> lsavailpprcport -dev IBM.2107-75ABTV1 -remotedev IBM.2107-7520781
-remotewwnn 5005076303FFC1A5 42:E4
IBM.2107-75ABTV1
=============================
I0011 I0400 FCP
I0012 I0402 FCP
3. Create the Remote Mirror and Copy paths between the two storage units using the
DS8000 CLI command mkpprcpath. In Example 4-34, paths are created between two
different ports on each storage unit, I0011:I0400 and I0012:I0402 (). These paths are
created using the same physical adapter on the local system. For more information, see
“Global Copy performance considerations” on page 46.
Example 4-34 Create pprc paths with mkpprcpath command

dscli> mkpprcpath -dev IBM.2107-75ABTV1 -remotedev IBM.2107-7520781 -remotewwnn
5005076303FFC1A5 -srclss 42 -tgtlss E4 I0011:I0400 I0012:I0402
IBM.2107-75ABTV1
CMUC00149I mkpprcpath: Remote Mirror and Copy path 42:E4 successfully
established.
Important: Remote Mirror and Copy paths need to be created between each LSS on
the local and remote storage units for use in the data migration.

4. Verify that the paths were created correctly using the lspprcpath command as shown in
Example 4-35 to confirm the path configurations. This command is issued against the
local LSS 01. The output from this command lists the two paths you created in step 2.
Example 4-35 Query the Remote Mirror and Copy paths with lspprcpath
dscli> lspprcpath -dev IBM.2107-75ABTV1 42
IBM.2107-75ABTV1
Src Tgt State SS Port Attached Port Tgt WWNN
=========================================================
42 E4 Success FFE4 I0011 I0400 5005076303FFC1A5
42 E4 Success FFE4 I0012 I0402 5005076303FFC1A5
4.4.3 Creating Remote Mirror and Copy pairs

Creating the Remote Mirror and Copy relationships between DS8000 storage units is the
same as between Enterprise Storage Server and DS8000. For more information, see 4.3.3,
“Creating Remote Mirror and Copy pairs” on page 64. Example 4-36 shows creating 64 global
copy pairs in interactive mode. The success messages from each pair are included for first
four relationships. The pairs can be created as Metro Mirror by substituting mmir for gcp when
-type is specified.
Example 4-36 Create Remote Mirror and Copy pairs using mkpprc
dscli> mkpprc -dev IBM.2107-75ABTV1 -remotedev IBM.2107-7520781 -type gcp -mode
full 4204-4243:E404-E443
IBM.2107-75ABTV1
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 4204:E404
4.4.4 Completing the data migration from DS8000 to DS8000

Moving data from one DS8000 to another DS8000 is the same as moving data from
Enterprise Storage Server to DS8000. For more information, see 4.3.4, “Completing the data
migration” on page 72. The only difference is that the source and target DS8000 are
displayed with all the information in the DS Storage Manager windows.

Figure 4-41 shows an example with three storage images available in the Select storage
complex list: ATS_20780, DS8k_TIC06v1_ATS and DS8k_TIC06v2_SLE. The last two storage
images belong to the same storage complex, which is logically partitioned (LPAR).
Figure 4-41 DS storage complex with two DS8000s
Figure 4-42 shows the new DS Storage Manager GUI.
Figure 4-42 Mirroring Connectivity view from release 6.1
Perform the following steps to complete the data migration using IBM Copy Services:
1. Monitor the copy progress using the lspprc -l command. The -l option is used to view
the out-of-sync tracks. The command and output are displayed in Example 4-37.
This output is formatted to highlight important information, specifically the out-of-sync
tracks and the First Pass Status.
Example 4-37 Query the status of the Remote Mirror and Copy pairs using lspprc -l
dscli> lspprc -dev IBM.2107-75ABTV1 4208-4243
IBM.2107-75ABTV1
==================================================================================
4208:E408 Copy Pending - Global Copy 100892 ... ... ... ... False

4209:E409 Copy Pending - Global Copy 114000 ... ... ... ... False
420A:E40A Copy Pending - Global Copy 114000 ... ... ... ... False
420B:E40B Copy Pending - Global Copy 108849 ... ... ... ... False
420C:E40C Copy Pending - Global Copy 114000 ... ... ... ... False
420D:E40D Copy Pending - Global Copy 114000 ... ... ... ... False
Query the state of the relationships until all pairs have zero (or near-zero) out-of-sync
tracks and the First Pass Status is True (Example 4-38).
Example 4-38 Monitor for wanted state

IBM.2107-75ABTV1
==================================================================================
4204:E404 Copy Pending - Global Copy 0 ... True
4205:E405 Copy Pending - Global Copy 0 ... ... ... ... .... True
2. After this state is reached, the application is stopped and all data is written to disk. Convert
the Global Copy to a synchronous copy by issuing the mkpprc command using mmir as the
type as shown in Example 4-39.
Example 4-39 Go-to-sync with mkpprc

dscli> mkpprc -dev IBM.2107-75ABTV1 -remotedev IBM.2107-7520781 -type mmir
4204-4207:E404-E407
IBM.2107-75ABTV1
3. Monitor the state of the relationships using the lspprc -l command until all pairs reach a
full duplex state. Notice in Example 4-40 that First Pass Status is changed to Invalid,
which is the expected output for a Metro Mirror relationship. This example shows both Copy
Pending and Full Duplex states.
Example 4-40 Pairs in Copy Pending and Full Duplex states

dscli> lspprc -dev IBM.2107-75ABTV1 -remotedev IBM.2107-7520781 -l
4204-4207:E404-E407
IBM.2107-75ABTV1
ID State Type Out Of Sync Tracks First Pass Status
=========================================================================
4204:E404 Copy Pending Metro Mirror 0 Invalid
4207:E407 Full Duplex Metro Mirror 0 Invalid

Continue to monitor the state until all relationships reach the Full Duplex state as shown
in Example 4-41.
Example 4-41 Check the state of the pairs with lspprc

IBM.2107-75ABTV1
===============================================================================
4204:E404 Full Duplex - Metro Mirror 0 ... ... ... ... .... Invalid
4205:E405 Full Duplex - Metro Mirror 0 ... ... ... ... .... Invalid
4206:E406 Full Duplex - Metro Mirror 0 ... Invalid
4207:E407 Full Duplex - Metro Mirror 0 ... Invalid
4. After the copy is complete, remove the Remote Mirror and Copy relationships using
rmpprc command. Example 4-42 shows removing the 64 relationships. You can use the
-quiet option as shown to disable the confirmation prompt.
Example 4-42 Remove Remote Mirror and Copy relationships

dscli> rmpprc -quiet -dev IBM.2107-75ABTV1 -remotedev IBM.2107-7520781
4204-4243:E404-E443
IBM.2107-75ABTV1
CMUC00155I rmpprc: Remote Mirror and Copy volume pair 4204:E404 relationship
5. Query the relationships to confirm the removal by issuing the lspprccommand. If the
relationships are all removed, the command indicates that no Remote Mirror and Copy
relationships were found as shown in Example 4-43.
Example 4-43 Confirm Remote Mirror and Copy relationships are removed
dscli> lspprc -dev IBM.2107-75ABTV1 -l 4204-4243
IBM.2107-75ABTV1
CMUC00234I lspprc: No Remote Mirror and Copy found.
If many LSSs are being used in the data migration, you might want to create a DS8000
CLI script to create paths and pairs. Automation allows you to verify all the information you
are passing to DS8000 CLI in a single place, and avoid any typing mistakes. A script can
also be used to set up queries to the pairs. These queries will let you know when to
synchronize Global Copy, or remove the pairs when full duplex state is reached.

Figure 4-43 provides a sample script that starts the go-to-sync function for Global Copy.
This example runs the command on 16 LSSs with 64 volumes each. The remote LSSs are
spread across two ranges: C8-CF and E8-EF. The script can easily be modified for new
scripts to query the pprc and remove the pairs.
#script to establish execute the go-to-sync function

mkpprc -dev IBM.2107-75ABTV1 -remotedev IBM.2107-7520781 -type mmir
0000-003f:c800-c83f
0100-013f:c900-c93f
0200-023f:ca00-ca3f
0300-033f:cb00-cb3f
0400-043f:cc00-cc3f
0500-053f:cd00-cd3f
0600-063f:ce00-ce3f
0700-073f:cf00-cf3f
0800-083f:e800-e83f
0900-093f:e900-e93f
0a00-0a3f:ea00-ea3f
0b00-0b3f:eb00-eb3f
0c00-0c3f:ec00-ec3f
0d00-0d3f:ed00-ed3f
0e00-0e3f:ee00-ee3f
0f00-0f3f:ef00-ef3f
Figure 4-43 Example of DS8000 CLI script
Start the script, using the -hmc option with the name or IP address of the HMC on the
primary system as shown in Example 4-44. You can also have a profile file set up. Place
the name of the script after -script option followed by the -user and -passwd options. In
this case, it is placed from a Windows XP Professional system,
Example 4-44 Running a DS8000 CLI script

C:\Program Files\ibm\dscli>dscli -hmc 9.155.62.102 -script gotosync.txt -user data
-passwd whateveritis
The data migration using IBM Copy Services is now complete.

4.5 Post-migration tasks
The final steps for accessing the data after the IBM Copy Services data migration is complete
vary by operating system. See the appropriate section for more details:
򐂰 AIX post-migration tasks
򐂰 Solaris and VxVM post-migration tasks
򐂰 System z post-migration tasks
4.5.1 AIX post-migration tasks

This subject is documented in Appendix A of IBM System Storage DS8000 Series: Copy
Services in Open Environments, SG24-6788, which is available at:
http://www.redbooks.ibm.com/redbooks/pdfs/sg246788.pdf
The recreatevg command is covered in the Information Center:

http://publib.boulder.ibm.com/infocenter/pseries/v5r3/index.jsp?topic=/com.ibm.aix
.cmds/doc/aixcmds4/recreatevg.htm
This section addresses considerations and clarification details not covered in those
references.
There are two phases to accessing an AIX volume group (VG) after an IBM Copy Services
copy operation:
1. Configuring the LUNs. Depending on the multi-path code, the LUNS can be hdisks,
vpaths, or something else if non-IBM storage is used.
This step is accomplished by assigning the LUNs to a host from the storage side, then
running cfgmgr at the host.
2. Accessing the VG.
Import the VG with importvg. You can also run recreatevg against the LUNs if another
copy (which can be the original VG) of the VG is located on the host.
Use recreatevg if a copy of the VG is already on the host because the disks will have
duplicate physical volume IDs (PVIDs), logical volume names, and file system mount
points. The AIX Logical Volume Manager (LVM) does not allow duplicates of these
characteristics. AIX does allow duplicate PVIDs on a system, but only as a temporary
situation.
If there is a LUN varied on in a VG with PVID, you cannot configure another hdisk/vpath on
the system with the same PVID using cfgmgr. To configure the disks in such a situation, you
must vary off the VG with the duplicate PVID, and then run cfgmgr.
You can also configure the LUN on AIX before the copy of data is placed on the LUN (which
creates the duplicate PVID). After the LUN is configured on AIX, subsequent FlashCopies or
Global Copies will not require the LUN to be configured again. The LUN will not have a PVID
because it has just been created.
Tip: It is not necessary to issue the following commands for clearing and setting a new
PVID on the LUNs:
# chdev -l <hdisk#> -a pv=clear
# chdev -l <hdisk#> -a pv=yes
These commands are run by the recreatevg command, so they can be skipped.

In the following example, you have an existing VG existingvg defined on the system, and want
to create a FlashCopy of it called flashcopyvg. To create the FlashCopy, perform the following
steps:
1. Create target LUNs on the storage and assign them to the host.
2. Configure the LUNs on the host with # cfgmgr (assuming you are using SDDPCM on a
DS8000, which results in one hdisk for each LUN).
3. Make a note of the new hdisks (in the example they are hdisk10, hdisk11, and hdisk12).
4. Initiate the FlashCopy. If the file systems are not unmounted, see “Maintaining file system
and data consistency” on page 91.
5. Run recreatevg to clean up the duplicate PVIDs, LV names, and so on, as follows:
# recreatevg -y flashcopyvg hdisk10 hdisk11 hdisk12
6. Vary on the VG with # varyonvg flashcopyvg.
Note: In this example, the new hdisks were configured before creating the FlashCopy,
so they do not have a PVID. However, if you perform the FlashCopy first, you need to
vary off existingvg before step 2.
If you want to update the copies later to match the VG again, follow these steps:
1. Vary off flashcopyvg using # varyoffvg flashcopyvg.
2. Export flashcopyvg with # exportvg flashcopyvg. This command removes information
about flashcopyvg from the ODM on AIX, but does not change any information on the
disks in the VG.
3. Create the copy with FlashCopy. For more information about when the application is not
quiesced and the file systems are not unmounted, see “Maintaining file system and data
consistency” on page 91.
4. Run recreatevg to create a VG from the FlashCopy volumes with new LV names using the
format # recreatevg -y flashcopyvg hdisk10 hdisk11 hdisk12. In this example, it is
called flashcopyvg This command also changes the PVIDs of the disks to unique PVIDs
and loads the ODM with this information.
Tip: You do not need to run importvg because the recreatevg command loads the
ODM with the VG information.
For more information, see the recreatevg man page at:

http://publib.boulder.ibm.com/infocenter/pseries/v5r3/index.jsp?topic=/com.ibm.
aix.cmds/doc/aixcmds4/recreatevg.htm
5. Vary on the VG using # varyonvg flashcopyvg.
Remember: In this example, step 2 is necessary because you are going to rerun
recreatevg, and you cannot have the VG already defined in the ODM.
You can make multiple copies of the VG using these steps.
Maintaining file system and data consistency

To maintain file system and data consistency, unmount the file systems because otherwise,
data might be in file system cache and not be written to disk.

If the application is running, make sure the FlashCopy (or whatever disk subsystem mirroring
method is used) data is a point-in-time image, and is consistent. Ensuring consistency
requires an application that is written to recover in the event of a system crash. If the
application does not support recovery, you must stop the application before initiating the
FlashCopy. In this case, unmount the file systems before the FlashCopy to flush the file
system cache to the file systems.
To create a FlashCopy of VGs used by a running application that supports recovery after a
system crash, perform the following steps:
1. Put the application in a hot backup or quiesced mode if the application supports it. This
speeds recovery of the application after it is started on the FlashCopy.
2. Use disk subsystem consistency groups for the disks in the VG.
3. Freeze the JFS2 file systems by using the chfs command if using Journaled File System 2
(JFS2). JFS2 is preferable because there is no similar function in JFS. For more
information, see the chfs man page at:
http://publib.boulder.ibm.com/infocenter/pseries/v5r3/index.jsp?topic=/com.ibm.
aix.cmds/doc/aixcmds1/chfs.htm
4. Preferably, have one file system log per file system.
5. Initiate the FlashCopy.
6. Thaw the JFS2 file system.
7. Turn off the hot backup or quiesce mode of the application.
8. Use these procedures to get the new VG activated on the system:
a. Run logredo against the file systems.
b. Run fsck against the file systems.
9. Verify the consistency of the data using the application that uses that data.
For more information about ensuring file system and data consistency, see:
http://www-1.ibm.com/support/docview.wss?rs=0&q1=%2bFlashcopy+%2bconsistency&uid=i
sg3T1000673&loc=en_US&cs=utf-8&cc=us&lang=en
4.5.2 Solaris and VxVM post-migration tasks

This section addresses the steps required to access the migrated SUN data after the IBM
Copy Services data migration is completed.
Moving the application to the DS8000

The steps for moving the application to the DS8000 are as follows:
1. Before you can split the Remote Mirror and Copy pairs, you need to stop any running
application.
2. Unmount the file system using the umount command, and verify that the file system is
unmounted using the df command as shown in Figure 4-44 on page 93.

# umount /ESS0
# df -k
Filesystem kbytes used avail capacity Mounted on
/dev/dsk/c1t0d0s0 33279289 5697404 27249093 18% /
/devices 0 0 0 0% /devices
ctfs 0 0 0 0% /system/contract
proc 0 0 0 0% /proc
mnttab 0 0 0 0% /etc/mnttab
swap 2736240 1296 2734944 1% /etc/svc/volatile
objfs 0 0 0 0% /system/object
/platform/sun4u-us3/lib/libc_psr/libc_psr_hwcap1.so.1
33279289 5697404 27249093 18%
/platform/sun4u-us3/lib/libc_psr.so.1
/platform/sun4u-us3/lib/sparcv9/libc_psr/libc_psr_hwcap1.so.1
33279289 5697404 27249093 18%
/platform/sun4u-us3/lib/sparcv9/libc_psr.so.1
fd 0 0 0 0% /dev/fd
swap 2735008 64 2734944 1% /tmp
swap 2734992 48 2734944 1% /var/run
swap 2734944 0 2734944 0% /dev/vx/dmp
swap 2734944 0 2734944 0% /dev/vx/rdmp
Figure 4-44 Unmount the file system and verify the results of that operation
3. Stop the VxVM volume by entering vxvol -g dg0 stop vol0.

4. Deport the volume group by entering vxdg deport dg0.
5. Issue the vxdisk offline IBM_SHARK0_0 command to VxVM offline the source LUN.
6. Verify the result of our activity by entering vxdisk list.
These last four steps are shown in Figure 4-45.
# vxvol -g dg0 stop vol0

#
# vxdg deport dg0
# vxdisk offline IBM_SHARK0_0
# vxdisk list
DEVICE TYPE DISK GROUP STATUS
IBM_DS8x000_0 auto:none - - online invalid
IBM_SHARK0_0 auto - - offline
c1t0d0s2 auto:none - - online invalid
Figure 4-45 VxVM-specific operations to remove the source LUN from the control of VxVM

7. Look at the disks that are visible by Solaris by entering the format command as shown in
Figure 4-46.
# format
Searching for disks...done
c2t500604843E0C4BCBd0: configured with capacity of 6.56MB

c4t500604843E0C4BD4d0: configured with capacity of 6.56MB
AVAILABLE DISK SELECTIONS:

0. c1t0d0 <SUN36G cyl 24620 alt 2 hd 27 sec 107>
/pci@8,600000/SUNW,qlc@4/fp@0,0/ssd@w21000004cfa37b97,0
/pci@8,600000/SUNW,qlc@4/fp@0,0/ssd@w21000004cfac3616,0
6. c2t50050763050C0786d4 <IBM-2107900-.437 cyl 4316 alt 2 hd 64 sec 256>
/pci@8,600000/fibre-channel@1/fp@0,0/ssd@w50050763050c0786,4
9. c2t5005076300C39629d0 <IBM-2105800-4.45 cyl 4313 alt 2 hd 64 sec 256>
17. c4t5005076300CD9629d0 <IBM-2105800-4.45 cyl 4313 alt 2 hd 64 sec 256>
/pci@8,700000/fibre-channel@1/fp@0,0/ssd@w5005076300cd9629,0
Specify disk (enter its number):
Figure 4-46 Using the format command to display the LUNs that are currently visible
Look for the source LUN paths c4t5005076300CD9629d0s2 and c2t5005076300C39629d0s2.

8. Split the Remote Mirror and Copy mirror and enable the zones for the DS8000 on the
switch so that the target LUN becomes visible.

9. Enter the format command as shown in Figure 4-47.
# format


6. c2t50050763030B048Ed0 <IBM-2105800-4.45 cyl 4313 alt 2 hd 64 sec 256>
/pci@8,600000/fibre-channel@1/fp@0,0/ssd@w50050763030b048e,0
Figure 4-47 Using the format command to display the LUNs that are currently visible
Comparing the output in Figure 4-46 on page 94 and Figure 4-47, you can see that you
now have two more disks. Actually, they are actually two paths to the same target (the
dual-pathed target device that you just made available to the operating system).
The output of the format command shows as an IBM 2105800. It is, however, a DS8000.
The DS8000 is labeled incorrectly because the target LUN is an exact replica of the
source Enterprise Storage Server LUN as created by Remote Mirror and Copy. The
source Enterprise Storage Server LUN initially was correctly labeled as IBM 2105800 by
Solaris, so the label (including the source geometry) was copied. However, do not rewrite
the label on the target LUN because user data might be affected.

You can, however, show that the target disk is a DS8000 using the following steps:
a. Issue the format command as shown in Figure 4-48.
# format


Specify disk (enter its number): 6
Figure 4-48 Using the format command to list disks

b. Enter the number of the disk you want to investigate. The Format menu for that disk
displays (Figure 4-49).
selecting c2t50050763030B048Ed0
[disk formatted]
FORMAT MENU:
disk - select a disk
type - select (define) a disk type
partition - select (define) a partition table
current - describe the current disk
format - format and analyze the disk
repair - repair a defective sector
label - write label to the disk
analyze - surface analysis
defect - defect list management
backup - search for backup labels
verify - read and display labels
save - save new disk/partition definitions
inquiry - show vendor, product and revision
volname - set 8-character volume name
!<cmd> - execute <cmd>, then return
quit
format> inq
Vendor: IBM
Product: 2107900
Revision: .437
Figure 4-49 Using the format menu to display inquiry data
c. Enter vxdisk list to see the status of the target volume. Figure 4-50 shows that VxVM
recognizes it as DS8000, despite the label on the disk that it is an IBM 2105-800. It
also shows that it is attributed with a flag udid_mismatch. This flag shows that VxVM is
aware that it is a cloned volume (Remote Mirror and Copy).
# vxdisk list
EMC0_0 auto:none - - online invalid
IBM_DS8x001_0 auto:cdsdisk - - online udid_mismatch
Figure 4-50 Using vxdisk list to show the status of our target device

10.Import the disk group by entering vxdg -o useclonedev=on -o updateid import dg0
(Figure 4-51).
# vxdg -o useclonedev=on -o updateid import dg0

# vxdisk list
IBM_DS8x001_0 auto:cdsdisk dg001 dg0 online clone_disk
Figure 4-51 Importing the disk group
11.Issue the vxprint command to see that the disk group was imported, but the VxVM
volume itself is still disabled (Figure 4-52).
# vxprint
Disk group: dg0
TY NAME ASSOC KSTATE LENGTH PLOFFS STATE TUTIL0 PUTIL0

dg dg0 dg0 - - - - - -
dm dg001 IBM_DS8x001_0 - 70598400 - - - -
v vol0 fsgen DISABLED 70596608 - CLEAN - -

pl vol0-01 vol0 DISABLED 70596608 - CLEAN - -
sd dg001-01 vol0-01 ENABLED 70596608 0 - - -
Figure 4-52 Entering vxprint to see volume status
12.To be able to mount it later on, start the volume by entering vxvol -g dg0 start vol0 as
# vxvol -g dg0 start vol0

# vxprint
Disk group: dg0

dg dg0 dg0 - - - - - -
dm dg001 IBM_DS8x001_0 - 70598400 - - - -
v vol0 fsgen ENABLED 70596608 - ACTIVE - -

pl vol0-01 vol0 ENABLED 70596608 - ACTIVE - -
sd dg001-01 vol0-01 ENABLED 70596608 0 - - -
Figure 4-53 Starting the volume

13.Mount the volume using the mount /ESS0 command (Figure 4-54).
# mount /ESS0
# df -k
/dev/dsk/c1t0d0s0 33279289 5697589 27248908 18% /
/devices 0 0 0 0% /devices
ctfs 0 0 0 0% /system/contract
proc 0 0 0 0% /proc
mnttab 0 0 0 0% /etc/mnttab
swap 2731024 1296 2729728 1% /etc/svc/volatile
objfs 0 0 0 0% /system/object
/platform/sun4u-us3/lib/libc_psr/libc_psr_hwcap1.so.1
33279289 5697589 27248908 18%
/platform/sun4u-us3/lib/libc_psr.so.1
/platform/sun4u-us3/lib/sparcv9/libc_psr/libc_psr_hwcap1.so.1
33279289 5697589 27248908 18%
/platform/sun4u-us3/lib/sparcv9/libc_psr.so.1
fd 0 0 0 0% /dev/fd
swap 2729792 64 2729728 1% /tmp
swap 2729776 48 2729728 1% /var/run
swap 2729728 0 2729728 0% /dev/vx/dmp
swap 2729728 0 2729728 0% /dev/vx/rdmp
/dev/vx/dsk/dg0/vol0 35298304 17205629 16961890 51% /ESS0
Figure 4-54 Mount the volume and verify that it became mounted
To verify the operation, check that the last entry in a file from the source LUN can also be
seen on the target LUN (Figure 4-55).
# cd /ESS0
# tail -3 samplefile
Mar 26 17:14:06 SunFire280Rtic2 Corrupt label; wrong magic number
Mar 26 17:14:06 SunFire280Rtic2 scsi: [ID 107833 kern.warning] WARNING:
/pci@8,700
************** LAST ENTRY TO SAMPLEFILE ******************
Figure 4-55 Check data on target LUN
4.5.3 System z post-migration tasks

There are two scenarios to consider when completing the data migration: Migrating system
DASD and migrating data. This section addresses at a high level the scenario of moving only
data. The System z support personnel know best how to move to the new DS8000 storage
unit. Use the following steps to move the data:
򐂰 Stop the applications.
򐂰 Delete the Remote Mirror and Copy pairs.
򐂰 Create new device entries in the input/output configuration data set (IOCDS) for the new
DS8000.
򐂰 Activate the new IOCDS with the new DS8000 devices online at IPL and the Enterprise
Storage Server devices offline at IPL. Also, ensure that future IPLs will use only DS8000
devices.

Stopping the applications
This step can be performed during a planned downtime of the system. All applications must
be stopped so they can be moved to the new storage. This example assumes that software to
automate the migration such as IBM GDPS® is not in place.
To stop the application successfully, perform the following steps:

1. Use the TSO command tso cquery devin(e000) bitmap as shown in Figure 4-56. The
output shows that the application is still running and there are still out-of-sync tracks (168).
The output also lists the four Remote Mirror and Copy paths available for this device. The
source volume is E000 and the remote volume is E008. The current state of the pair is
PENDING.XD.
CQUERY FORMATTED LVL 4

VOLUME REPORT
*------ --------- ---------- ----------- --------- --------- *
* E000 PRIMARY.. PENDING.XD ACTIVE.. E000 00 00 E008 00 08 *
* CRIT(NO)....... CGRPLB(YES) 0000000FNZT1 0000000FNZT1*
* INCRES(NO). *
* ----- --------- ------ ------------------- *
* IF STATE = PENDING/SUSPEND: TRACKS OUT OF SYNC = 168 *
* TRACKS ON VOLUME = 114000 *
***
* ----------- ---------------- ----------- *
* PRIMARY.... 5005076305FFC6F3 5.2.420.266 *
* SECONDARY.1 5005076305FFC6F3 *
********************************************************************
CQUERY COMMAND COMPLETED FOR DEVICE E000. COMPLETION CODE: 00
***
Figure 4-56 TSO query of global copy pair
2. Stop the application.

3. Issue the tso cquery devin(e000) bitmap command again. The output shown in
Figure 4-57 shows that the copy is now at 100%.

VOLUME REPORT
*------ --------- ---------- ----------- --------- --------- *
* E000 PRIMARY.. PENDING.XD ACTIVE.. E000 00 00 E008 00 08 *
* INCRES(NO). *
* ----- --------- ------ ------------------- *
* ----------- ---------------- ----------- *
***
* PRIMARY.... 5005076305FFC6F3 5.2.420.266 *
********************************************************************
***
Figure 4-57 Global copy pair status
4. Issue the go-to-sync command and verify that the pair is in duplex state as shown in
Figure 4-58.

VOLUME REPORT
*------ --------- ---------- ----------- --------- --------- *
* E000 PRIMARY.. DUPLEX.... ACTIVE.. E000 00 00 E008 00 08 *
* INCRES(NO). *
* ----- --------- ------ ------------------- *
* ----------- ---------------- ----------- *
* PRIMARY.... 5005076305FFC6F3 5.2.420.266 *
***
********************************************************************
***
Figure 4-58 Verify duplex state

Deleting the Remote Mirror and Copy pairs
For information about the TSO and ICKDSF commands to delete the Remote Mirror and
Copy pairs, see 4.3.4, “Completing the data migration” on page 72. Using the TSO command
cquery again shows that the volume is now in simplex state and all the paths are deleted
(Figure 4-59).

VOLUME REPORT
*------ --------- ---------- ----------- --------- --------- *
* E000 ......... SIMPLEX... INACTIVE E000 00 10 .......... *
* ............... ........... 0000000FNZT1 ............*
* ........... *
* PATHS SAID DEST STATUS: DESCRIPTION *
* ----- --------- ------ ------------------- *
* 0 ---- ---- 00 NO PATH............ *
* ---- ---- 00 NO PATH............ *
* ---- ---- 00 NO PATH............ *
* ---- ---- 00 NO PATH............ *
* ----------- ---------------- ----------- *
* PRIMARY.... 5005076305FFC6F3 5.2.420.266 *
***
********************************************************************
***
Figure 4-59 Volume in simplex state and paths removed
Creating IOCDS entries

Perform this step in advance to decrease the downtime for the application. The system
support personnel make the changes required on the system to recognize the new storage
serial number and devices.
New DS8000 devices

Activate the new IOCDS so that the new DS8000 devices are online and the old storage
devices offline at IPL. This change allows the system to use only the new DS8000 devices on
any future IPLs.
The new DS8000 devices are varied online and available for use. The application is ready to
restart.
5
Chapter 5. DSCLIbroker
The DSCLIbroker is a scripting framework that automates Copy Services functions. The
DSCLIbroker provides the following features:
򐂰 Grouping volumes according to applications or other context
򐂰 Simplifies the execution of Copy Services commands
򐂰 Provides a scripting framework for implementing automation functions

򐂰 Overview
򐂰 System environment
򐂰 Programming techniques
򐂰 Automation techniques
򐂰 Considerations about configuration data
򐂰 Example migration using DSCLIbroker

5.1 Overview
Besides the storage capacity for application data, the storage infrastructure of a modern data
center must also replicate the data to other storage devices. With the Copy Services and
interfaces like DSCLI, applications can use this replication by controlling Copy Services for
their own purposes. These functions are commonly used for disaster prevention or data
backup purposes.
Copy Services are frequently used for data migrations, but migrations are not a daily
operation. Depending on your data center environment, applications, and type of migration,
the migration tasks can get complicated. As a result, you might want to automate certain
tasks, especially when the required actions cannot be accomplished using the standard
storage management software.
The DSCLIbroker is a scripting framework that allows you to create user customized
automation scripts. For example, consider a multitiered layered stack consisting of the
DS8000 hardware as the lowest level and DSCLI above. You might want to write automation
scripts where DSCLI commands are run against the storage. The DSCLIbroker can be
positioned as an extra layer between the DSCLI and the applications (Figure 5-1).
A p p lic a tio n s
T P C -R C u s to m e r
w r itte n s c r ip ts D S C L Ib r o k e r
D SC LI
H a r d w a re (D S 8 0 0 0 )
Figure 5-1 Positioning of DSCLIbroker
The framework is written in Perl scripting language and consists of a series of perl library
modules that support all DS8000 Copy Services functions. You can write your own scripts
using the library. The framework also provides a perl script for each Copy Services DSCLI
command that can be used without additional programming.
DSCLIbroker is available as an STG-Lab Service. Contact your IBM representative to order

this service.
In the subject field, enter Migration using DSCLIbroker and continue to process the form.
If you are outside of IBM, talk to your IBM Representative about ordering the Service.
5.1.1 DSCLIbroker concepts

The complexity of Copy Service configuration increases as the number of applications
involved and the size of those applications increases. The more applications must be
migrated at the same time, the more care must be taken to map the correct volumes to the

applications. In addition, larger application can have large lists of volumes that have copy
relations to other volumes. Migrating data to the correct volumes is vital to avoid data
inconsistency.
Using DSCLI Copy Services commands, with every command a list of copy relations must be
specified. In a migration scenario with the steps: mkpprc, lspprc, pausepprc, lspprc,
failoverpprc and again lspprc, the list of copy relations must be specified six times. It
requires much effort to maintain these commands either in self written scripts or on the
DSCLI command line.
With DSCLIbroker, the configuration data of the Copy Services relations is separated from the
scripting code in a repository. In this repository, multiple Copy Services relations belonging to
a single application can be grouped together and tagged with a name. When a DSCLIbroker
script is run, it refers to the tagged name, and the DSCLIbroker then fetches all copy relations
from the repository. Maintenance of these relationships is done in the repository, and so you
do not need to change the scripting code.
In addition, the DSCLIbroker provides a scripting framework that offers an easy way to write
user customized scripts. This framework is implemented in modular libraries written in perl
scripting language with an object-oriented approach. There is one library for each DS8000
Copy Services function available. The libraries themselves are organized so they can be
extended to support other storage platforms too. The current storage platforms supported
include DS8000, IBM DS6000™ and ESS Model 800. Plans are in place to support SAN
Volume Controller and V7000 storage platforms and the TotalStorage Productivity Center for
Replication command-line interface.
DSCLIbroker libraries
The libraries are the core of the scripting framework. If you write perl scripts using the
DSCLIbroker, you must include the libraries.
The following are the libraries and a short summary of their purposes and contents:
DSPPRC.pm This library is an object class where all remote copy functions are
implemented. The functions include managing the paths and the pair
relations.
DSFlashCopy.pm This library is an object class that holds all functions that maintain
FlashCopy relations.
DSGlobalMirror.pm This library is an object class as well. It holds all functions related to
DS8000 Global Mirror.
DSCLInator.pm This library is the meta class for the preceding classes. Commonly
used functions for Copy Services are located here.
DSlib.pm This library contains global functions with no relations to Copy
Services functions such as maintaining the DSCLIbroker environment
and querying and retrieving data from the repository.
DBbox.pm This library contains all necessary functions to communicate with the
storage subsystems.
Chapter 5. DSCLIbroker 107

Figure 5-2 shows an overview of the libraries of the DSCLIbroker.
D SC LInator.pm
_ collect_option s
_ xml_pa rse
_ xml_sho w
_ re ad_License
DS PP RC .pm
_c hec k_ directio n
_c ollect_pa ir s
mkpprcpath
rmppr cp ath
ls pp rcp ath
D Slib .p m D SFlashC o py.p m scanpp rc path
se tD ebug Level _co lle ct_p airs mkpprc
D Sbo x.pm g etD ebug Level DS GlobalM irror.pm scan _flash pausp prc
se tBox se tS imulate Mod e mk g mir mkflas h rmppr c
g et_d evID g etSimulate Mo de pausgmir rmflash fa ilo verpprc
g et_w w nn g etC onfig r es ume gmir resyncfla sh fa ilb ackpprc
g et_b oxN ame q uer yStanza r mg mir reversefla sh re sumepprc
C reateD SC LIs cript g etSta nza show gmir revertflas h freeze ppr c
W riteD SC LIsc ript _ _chec k_ results mk sessio n commitflash unfre ezepprc
SetCommandMod e g et_ cfgF ile r ms es sion un freeze flash ls pp rc
D oD SC LIco mman d o penSe ssio n lss e ssion lsflash scanpp rc
Figure 5-2 DSCLIbroker libraries
The data repository

All configuration data is stored in the data repository, and is used by the DSCLIbroker libraries
when composing commands for the DSCLI. In this sense, configuration data includes the
following information:
򐂰 Copy relations defined by source volume to target volume.
򐂰 Copy paths information for remote copy, which includes the relation source LSS to target
LSS used by source port to target port.
򐂰 Definitions for DS8000 Global Mirror.
򐂰 Information about the storage systems.
Because some DSCLI commands use the same information, this information can be collected
in the same repository entity. For example, when creating a Metro Mirror or a Global Copy, the
same DSCLI command is used. Metro Mirror is denoted by the option -type mmir and Global
Copy uses the option -type gcp. The remaining parameters are the source and target device,
and the source and target volumes. A FlashCopy needs the same set of information except
that the target device is not required. All these copy pair relations can be described with the
same set of data, which is in the form of a database table or stanza file.

The complete set of required information is shown in Figure 5-3.
CopyPairs.cfg DSdev.cfg
name name
T ype IPaddress1
PPRCpath.cfg
[mmir|gcp|flash] IPaddress2 name
Source WWNN srclss
Target DSCLIprof tglss
srcvol pwfile srcport
tgtvol user tgtport
optset Session.cfg Source
Target
name
consist
GlobalMirror
PPRCpairs
LSS
SessionID
Volumes
OptSet.cfg
name
param
GlobalMirror.cfg
value name
Box
SessionID
subordpath
OptSet
LSS
Figure 5-3 DSCLIbroker data model
This set of tables is a normalized data model, which in theory can be used by a relational
database system. For DSCLIbroker however, the tables are implemented as flat stanza files
because they can handle hundreds of thousands of entities without any problems. When the
data migrations are done, the data in the repository is obsolete and can be discarded. In other
engagements, the data in the repository must be maintained for a longer time. In these
engagements, the data management capabilities of a database system might be more useful
or even required.
As shown in Figure 5-3, each table has a name that is the key to a set of configuration data. A
table might have an entry that refers to a key name of another table. For example, in the table
CopyPairs.cfg, the entries for Source and Target reference a storage device defined in the
table DSdev.cfg.
The following section includes an example for a complete repository definition d where an
application named SAPHR1 manages a copy relation using the DSCLIbroker. The whole
setup includes the following steps:
1. Define the copy relations.
2. Define the storage devices.
3. Define the paths.

In the example, the application SAPHR1 is using the volumes 820A-820C and 8410-841F,
which are two different contiguous volume ranges. The complete copy relation for this
application is denoted in the stanza file CopyPairs.cfg. One stanza is created for the first
volume range, and another for the second volume range. Both stanzas have the same name:
SAPHR1_ab (Example 5-1).
Example 5-1 Defining a complex copy relation

CopyPair {
name = SAPHR1_ab
type = gcp
srcss = 82
tgtss = 8E
srcvol = 820A-820C
tgtvol = 8E0A-8E0C
source = ATS_3
target = ATS_1
seqnun =
optset = gcp_cascade
}
CopyPair {
name = SAPHR1_ab
type = gcp
srcss = 84
tgtss = 8E
srcvol = 8410-841F
tgtvol = 8E10-8E1F
source = ATS_3
target = ATS_1
seqnun =
optset = gcp_cascade
}
In some cases, one or more command-line options are required multiple times. For example,
a Global Copy relation that is cascaded to an existing remote copy relation requires the option
-cascade. This option can be placed in the repository using the stanza file OptSet.cfg as
shown in Example 5-2. The defined option set gcp_cascade is referenced in the CopyPair
stanza with the tag optset, as shown in Example 5-1.
Example 5-2 Define parameter in OptSet.cfg

OptSet {
name = gcp_cascade
param = cascade
value = yes
}
When this entry is referenced in a CopyPairs.cfg stanza, the option -cascade is the default
every time DSCLIbroker generates a DSCLI command for this relation.

In the CopyPair stanzas, the entries for source and target refer to the source and target
storage devices. Both entries need to be defined in the DSdev.cfg file. In Example 5-3, the
source device is named ATS_3 and the target device is named ATS_1.
Example 5-3 Definition of the storage devices

DSdev {
name = ATS_1
IPaddress1 = 9.155.70.26
IPaddesss2 = 9.155.70.55
WWNN = 5005076303FFC08F
devID = IBM.2107-7503461
DSCLIprof = script_03461.profile
PWfile = script_03461.pwfile
user = script
}
DSdev {
name = ATS_3
IPaddress1 = 9.155.62.97
IPaddesss2 = 9.155.62.97
WWNN = 5005076303FFC1A5
devID = IBM.2107-7520781
DSCLIprof = script_20781.profile
PWfile = script_20781.pwfile
user = script
}
Remember: As you can see in the stanza, password files are used for the authentication to
DSCLI. When writing scripts that runs command against the DSCLI, an automated
authentication to the DSCLI is useful. Otherwise you must type in the user name and
password each time the script is run. DSCLI offers a secure method to log in to the DSCLI
automatically. For more information, see the Command-Line Interface User’s Guide for the
DS6000 series and DS8000 series, GC53-1127, at:
http://www-05.ibm.com/e-business/linkweb/publications/servlet/pbi.wss
After identifying the storage devices, the paths must be specified. To establish the paths for
copy relations, a physical link must be established first (see 4.4.2, “Creating Remote Mirror
and Copy paths” on page 83). Verify that the links are available and the required information
for the stanza entries can be obtained using the script lsavailpprcport.pl, as shown in
Example 5-4.
Example 5-4 Obtaining the port information for the paths

$ lsavailpprcport.pl -source ATS_1 -target ATS_3 -srclss 84 -tgtlss 8e
==================================================
IBM.2107-7503461/I0141 IBM.2107-7520781/I0402 FCP
IBM.2107-7503461/I0142 IBM.2107-7520781/I0400 FCP
$_
In this example, two ports have physical connections to the target storage device. Use this
information to create the stanza file for the paths. For each repository stanza file, a script is

available that generates the stanza information automatically. Use these scripts for the path
definitions. The path definitions are the most vital in the repository because the configuration
of the path can get complex. Only when the path definitions are correct will the subsequent
copy commands work properly.
Use the gen_pprcpaths.pl command to generating path stanzas as shown in Example 5-5.
The option -name SAPHR1 defines the base name of the stanza. The option -d ‘f:ab’
specifies that the paths data are created for the forward direction only. This option adds _ab to
the stanza base name, which results in the real stanza name SAPHR1_ab. The option -l is
used to specify the LSS relation as given. The option -p is used for the port pairs as shown by
the lsavailpprc.pl command in Example 5-4 on page 111.
Example 5-5 Generating path stanzas

$ gen_pprcpaths.pl -name SAPHR1 -d 'f:ab' -l '82:8E 84:8E' -p 'I0141:I0402
I0142:I0400' -s ATS_1 -t ATS_3
##################################################
#
# PPRC paths for SAPHR1_ab
#
##################################################
#############
# LSS 82 -> 8E
# Box ATS_1 -> ATS_3
#############
PPRCpath {
name = SAPHR1_ab
srclss = 82
tgtlss = 8E
srcport = I0141
tgtport = I0402
Source = ATS_1
Target = ATS_3
consist = no
}
PPRCpath {
name = SAPHR1_ab
srclss = 82
tgtlss = 8E
srcport = I0142
tgtport = I0400
Source = ATS_1
Target = ATS_3
consist = no
}
#############
# LSS 84 -> 8E
# Box ATS_1 -> ATS_3
#############
PPRCpath {
name = SAPHR1_ab
srclss = 84
tgtlss = 8E
srcport = I0141

tgtport = I0402
Source = ATS_1
Target = ATS_3
consist = no
}
PPRCpath {
name = SAPHR1_ab
srclss = 84
tgtlss = 8E
srcport = I0142
tgtport = I0400
Source = ATS_1
Target = ATS_3
consist = no
}
5.1.2 DSCLIbroker scripts

The DSCLIbroker comes with scripts that can be used as soon as the repository has the
required configuration data. For FlashCopy, Global Copy, and Metro Mirror, the stanzas for the
Storage device must contain the paths and the copy pairs. For more information, see 5.1.1,
“DSCLIbroker concepts” on page 106. For Global Mirror, the stanzas GlobaMirror.cfg and
Session.cfg must also be filled with data.
For each Copy Services-related DSCLI command, a corresponding DSCLIbroker script is

available. Each script provides all command options that you can use in DSCLI. Some scripts
have additional options that provide additional functionality. For example, using the -p option
with the command failoverpprc.pl pauses the relation at the primary site before the failover.
All scripts require the option -name, where the corresponding tagged name of the stanzas is
requested. If no options are provided, you see a help text where the complete syntax is
shown. Example 5-6 shows the help text for the command mkpprc.pl.
Example 5-6 Help output example

$ mkpprc.pl
Usage:
./mkpprc.pl
[-h|help]# This output
[-d|debug 1-4]# Set the debug level. Recommended level: 2
[-s|simulate]# Run script in simulate mode. Requires -debug 2
[-b|banner]# Prints out a banner
[-direction forward|reverse]
# Specifies in which direction of the pairs
# should operate. The default is -d forward.
[-t|type mmir|gcp]
# Overrides the type specified in the stanzas
[-m|mode full|nocp]
# Overrides the copy mode specified in the stanzas
[-cascade]# Enable relation to be cascaded
[-nocascade]# Overwrite the cascade option if set in OptSet
[-incresync enable|enableoinit|disable|recover|override]
# apply incremental resync. Overrides the stanzas
[-to|tgtonline]# enables target to get online (z/OS only)
[-tr|tgtread]# Allows read access from target volumes

[-suspend]# suspend relations after task has finished
[-crit|critmode]# enables critmode (z/OS only)
[-disableautoresync]# disable the auto resync functionality of Global Copy
[-rr|resetreserve]# reset the reservation on the target volumes
[-tgtse]# specifies the target volumes as space efficient
[-w|wait]# wait until initial copy has completed
# WATCH OUT: May take a while. Make sure that IPC_Timeout and NET_Timeout in
# DSCLIbroker.cfg matches. Otherwise the session may break down!!!
[-nooppt]# Will not use a defined option set
-n|name <CopyPairName>
# Specifies the name of the copy pairs
$_
The following examples are based on the configuration data that was created in the previous
chapter. You can now establish the Global Copy relation using the script mkpprc.pl as shown
in Example 5-7. The only required parameter you must specify is the name of the stanza
where all copy relations are defined. SAPHR1_ab is the corresponding stanza as shown in
Example 5-1 on page 110.
Example 5-7 Establish the Global Copy with mkpprc.pl

$ mkpprc.pl -n SAPHR1_ab
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 820A:8E0A successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 820B:8E0B successfully created.
... snippet ...
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 841F:8E1F successfully created.
$_
All other scripts work in the same way. The only required parameter is the name of the copy
relation located in the stanzas. Additional parameters can be supplied depending on what you
want to do. For an overview of the available options, use the option -help or run the command
without any parameters. Example 5-8 shows how to pause the Global Copy.
Example 5-8 Pause Global Copy with pausepprc.pl

$ pausepprc.pl -n teamblack_ab
CMUC00157I pausepprc: Remote Mirror and Copy volume pair relationship 820A:8E0A successfully paused.
CMUC00157I pausepprc: Remote Mirror and Copy volume pair relationship 820B:8E0B successfully paused.
... snippet ...
CMUC00157I pausepprc: Remote Mirror and Copy volume pair relationship 841F:8E1F successfully paused.
$_
In the previous examples, no information is provided about how the script is generating the
DSCLI commands. To make the scripts more verbose, the -d (debug) option can be used.
There are four levels of debug information available. The first debug level shows the
generated DSCLI command.
Another helpful option is -simulate, which displays the DSCLI command but does not run it.
This option can be used to verify whether the generated DSCLI command is the one you are
expecting before it takes effect. The option -d 1 also shows the generated command, but the
command is run.
Example 5-9 shows the output when using the debug and simulate options.
Example 5-9 Simulate mode and verbose command execution for mkpprc.pl
$ mkpprc.pl -n SAPHR1_ab -simulate

mkpprc -dev IBM.2107-75ABTV1 -remotedev IBM.2107-7503461 -mode full -type gcp 820A-820C:8E0A-8E0C
8410-841F:8E10-8E1F
$
$ mkpprc.pl -n SAPHR1_ab -d 1
Looking for name: SAPHR1_ab
mkpprc -dev IBM.2107-75ABTV1 -remotedev IBM.2107-7503461 -mode full -type gcp 820A-820C:8E0A-8E0C
8410-841F:8E10-8E1F
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 820A:8E0A successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 820B:8E0B successfully created.
... snippet ...
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 841F:8E1F successfully created.
$_
When a PPRC relation has to be failed over, the DSCLI command failoverpprc must be
issued at the target storage device. Provide the pair relations in the reverse order. The
failoverpprc.pl script does both automatically. In Example 5-10 the generated
failoverpprc command is displayed. Comparing it to the Example 5-9 on page 114, the
device IDs of the storage devices and the pair relations are reversed.
Example 5-10 Fail over Global Copy

failoverpprc.pl -n SAPHR1_ab -d 1
Looking for name: SAPHR1_ab
failoverpprc -dev IBM.2107-7503461 -remotedev IBM.2107-75ABTV1 -type mmir 8E0A-8E0C:820A-820C
8E10-8E1F:8410-841F
CMUC00196I failoverpprc: Remote Mirror and Copy pair 8E0A:820A successfully reversed.
CMUC00196I failoverpprc: Remote Mirror and Copy pair 8E0B:820B successfully reversed.
... snippet ...
CMUC00196I failoverpprc: Remote Mirror and Copy pair 8E1F:841F successfully reversed.
$_
5.1.3 User customized scripts

You can write your own scripts if you have basic perl scripting language skills. The perl scripts
must include the DSCLIbroker perl libraries as described in “DSCLIbroker libraries” on
page 107. The DSCLIbroker libraries are designed as a modular set of building blocks to
make creating scripts easier.
The following example demonstrates a simplified migration to target DS8000 storage devices
using a Global Copy replication. The following steps are processed by the scripts:
򐂰 Create the paths to the target storage device.
򐂰 Establish a Global Copy and wait until all tracks have been copied.
򐂰 Pause the Global Copy.
򐂰 Fail over Global Copy to the target site.
Using this script, you can perform a series of migrations without any changes to the code. For
each application that needs to be migrated, you must change the data in the corresponding
stanza files.
Example 5-11 shows the complete script. This script uses the libraries DSlib.pm, where the
debug levels and the simulation mode are included, and DSPPRC.pm, where all remote copy
functions are defined. The libraries are referenced in lines 6 and 7 of the script.
Example 5-11 Example migration script using DSCLIbroker
1 #!/usr/bin/perl -w
2

3 #
4 # Include required library modules
5 #
6 use DSlib;
7 use DSPPRC;
8
9 #
10 # Declare command line options
11 #
12 use Getopt::Long;
13 my ($opt_debug,
14 $opt_simulate,
15 $opt_name,
16 );
17
18 GetOptions(
19 "d|debug:i" => \$opt_debug, # --debug
20 "s|simulate" => \$opt_simulate, # --simulate
21 "n|name=s" => \$opt_name, # --name <stanza_name>
22 );
23
24 if (!defined $opt_name) {
25 print "option -name is required\n";
26 exit(3);
27 }
28
29 #
30 # set debug level and simulator mode
31 #
32 my $dbg_level=setDebugLevel($opt_debug);
33 my $dbg_simulate=setSimulateMode($opt_simulate);
34
35 #
36 # Define source and target device objects
37 #
38 my $source=DSbox->new();
39 my $target=DSbox->new();
40
41 #
42 # Create PPRC thingy
43 #
44 my $PPRC=DSPPRC->new($source,$target);
45
46 #
47 # Apply command for the primary site
48 #
49 $PPRC->mkpprcpath($opt_name);
50 $PPRC->mkpprc($opt_name,'forward','-wait');
51 $PPRC->pausepprc($opt_name);
52
53 #
54 # Execute primary DSCLI commands
55 #
56 my $ret=$source->DoDSCLIcommand('server');
57

58 #
59 # Apply failoverpprc at secondary site
60 #
61 $PPRC->failoverpprc($opt_name);
62 $ret=$target->DoDSCLIcommand('server');
63
64
65 exit($ret);
In the lines 9 - 27 of the script, the required command options are declared. The -name option
is required and therefore the lines 24 - 27 are checking whether this option is supplied. If the
option is not given, an error message is reported and the script exits.
Verify the DSCLI commands that are generated by this script before they take effect.
Therefore, the simulation mode is enabled when the command-line option -simulate is
specified. The simulation mode is engaged in line 33.
In lines 35 - 44, the necessary objects are defined. There is one object for each storage
device required, and an object that provides the remote copy function to the script.
In lines 49 - 51, the DSCLI commands for the paths and the copy pairs are generated. In this
case, a Metro Mirror is established with the option -wait. The Metro Mirror allows all tracks to
be copied to the secondary site before the operation continues. When the copying is
complete, the copy relation is paused using the pausepprc command. This command
sequence must be applied to the primary storage device.
In line 56, all DSCLI commands generated in the previous lines are sent to the primary
storage device for execution.
In line 59, the failover to the secondary device is applied in the same manner. This command
is sent to the auxiliary storage device. This script is now completed.
Example 5-12 shows the output of the simulation mode of the script, which is the generated
sequence of DSCLI commands. The commands, up to the pausepprc command, are sent to
the primary storage device. The failoverpprc command is sent to the target storage device.
Example 5-12 Simulation mode output

$ ITSOexamp.pl -name app1_ab -s
mkpprcpath -dev IBM.2107-75ABTV1 -remotedev IBM.2107-7503461 -remotewwnn
5005076303FFC08F -srclss 42 -tgtlss 64 -consistgrp I0011:I0142 I0012:I0141
mkpprcpath -dev IBM.2107-75ABTV1 -remotedev IBM.2107-7503461 -remotewwnn

5005076303FFC08F -srclss 53 -tgtlss 57 -consistgrp I0011:I0142 I0012:I0141
mkpprc -dev IBM.2107-75ABTV1 -remotedev IBM.2107-7503461 -mode full -type -wait

4200-4203:6400-6403 5300-5303:5700-5703
pausepprc -dev IBM.2107-75ABTV1 -remotedev IBM.2107-7503461 4200-4203:6400-6403

5300-5303:5700-5703
failoverpprc -dev IBM.2107-7503461 -remotedev IBM.2107-75ABTV1 -type mmir

6400-6403:4200-4203 5700-5703:5300-5303
$_

This script can be used for a series of migrations. In this example, the migrations are done
application by application. In other words, this script will migrate one application after the
other. For each migration, the same script is run, with the name of the application is supplied
with the option -name. The only thing that must be changed are the copy relations for each
application, which must be defined correctly in the CopyPairs.cfg stanza file.
5.1.4 Additional useful scripts

The DSCLIbroker framework includes additional scripts that help maintain the data repository.
There are also other useful scripts that enable enhanced functions for the DSCLI commands.
The following are the additional scripts and their functions:
򐂰 gen_dsdev.pl, gen_gmir.pl, gen_pprcpairs.pl, gen_pprcpaths.pl, gen_session.pl:
These scripts generate the stanza files that correspond to their name. They format the
stanza files and place the values you supply with the command file options in the correct
places. The scripts gen_pprcpaths.pl and gen_pprcpairs.pl are especially useful
because they allow you to list multiple relations in the command-line options. For more
information about creating stanzas dynamically, see 5.4, “Automation techniques” on
page 125.
򐂰 ValidatePairs.pl, ValidatePaths.pl: These scripts validate the pair and path
configuration. This means that the content of the data in the stanza is compared against
the real status on the storage devices.
򐂰 scanpprc.pl, scanflash.pl, scanpprcpath.pl, scansession.pl: The scan scripts obtain
information from the storage devices and the output from the DSCLI is sent back to the
DSCLIbroker as an XML data format. This information allows the scan commands to
select specific parameters and display their values in the window.
򐂰 QUERY.pl: This script allows you to retrieve selected data from the repository. QUERY.pl
searches after a pattern in a stanza file and prints out the results as a stanza or separated
by commas.
򐂰 lspprc.pl -sum, lsflash.pl -sum: The option -sum with the scripts lspprc.pl and
lsflash.pl prints the total number of Out-of-Sync tracks. This list is useful when many
copy pairs are involved. Otherwise, to obtain the total amount of Out-of-Sync tracks for
each single output line, the Out-of-Sync tracks must be summarized manually.
򐂰 lspprc.pl -waitnull, lsflash.pl -waitnull: The option -waitnull is similar to the -sum
option, but the command will not return until all tracks have been copied to the remote
target volumes.
5.2 System environment

The system environment of the DSCLIbroker is divided in a server part and a client part. In
this section, the architecture of the DSCLIbroker, and how the communication and the
maintenance are organized are described.
5.2.1 Architecture
The DSCLIbroker itself is organized as a client server application. The server part includes
the broker itself. The broker is a daemon that waits for a request from a client to open a
connection to a storage device. This connection is called a session. The session is opened by
forking a child process where the DS8000 command-line interface (DSCLI) is started. The
session waits at the input prompt. The broker manages the communication to the client.

The server where the DSCLI is running must have the role of a trusted storage management
server. The server hardware must have access to the DS8000 storage systems, but they must
also be accessible by storage administrators. This server can additionally be a gateway to the
storage environment. Therefore it must be equipped with two different network interfaces.
One interface is used for the administrator access and the second one is used to
communicate with the storage environment. Figure 5-4 shows the system architecture.
The server also hosts the repository with all its stanza files. In this way, the configuration data
is in one centralized spot and does not need to be distributed to the administrators.
Repository
Administrator 1
connects
Broker
connects
Administrator 2 starts
dscli dscli dscli

1 2 3
connects
Administrator n DSBox A DSBox B
Client Repository Server
Figure 5-4 DSCLIbroker architecture
The client part of the DSCLIbroker is dedicated to the storage administrators, who work with
one or more storage devices. Every DSCLIbroker script that runs at the client site
communicates with the server to perform the following tasks:
򐂰 Retrieve data from the repository
򐂰 Compose the DSCLI commands
򐂰 Send the commands back to the server
The server forwards the commands to the corresponding DSCLI daemon waiting in the
background. When the command is run by the DSCLI, the results are sent back to the client.
5.2.2 Communication
The communication between the DSCLIbroker client and the server is organized in sessions.
The storage administrator, working as a client, must establish a session to a storage device.
Use the script startSession.pl to send a request with a name to the corresponding entry in

the DSdev.cfg stanza file to the broker. When the broker finds the stanza entry, it uses this
information to fork a process call the worker and starts it in the DSCLI.
The broker sends the client a unique session ID and the port where the worker is listening.
This information is located in a dedicated session file at the client. The broker then sends a
lssi command to the DSCLI in the background. The output of this command is sent back to
the client. Use this output to verify whether the session was opened against the correct
DS8000 storage system.
A session can also be closed. The command stopSession.pl sends an appropriate request
to the corresponding worker process that quits the DSCLI and terminates the process.
5.2.3 Users
Multiple users can be defined in DSCLI for each storage system. Users can also be defined
using the DSCLIbroker. To do so, define an additional stanza in the DSdev.cfg stanza file for
the same storage system. The user name must be declared (Example 5-3 on page 111). The
session must be started with that user name supplied using the option -user as shown in
Example 5-13.
Example 5-13 Start session for a dedicated user.

$ startSession.pl -box ATS_3 -user admin03
-----------------------------------------
=================================================================================
ATS_20780 IBM.2107-7520781 IBM.2107-7520780 922 5005076303FFC1A5 Online Enabled
dscli>
-----------------------------------------
Session on port 55559 for ATS_3 with id 1311037799 was created successfully!
$_
Although with DSCLI a user can log in multiple times to the same storage device, with
DSCLIbroker only one session per storage device is allowed.
The authentication to the DSCLI is done using a password file, which must be generated
manually in DSCLI. The password files must be located in the directory ./pwfiles of the
home directory of the DSCLIbroker scripting framework. This password file is an encrypted
file, and access should be restricted to read access for the administrator only. The password
file is generated with the DSCLI command managepwfile. See the DSCLI online help or the
DSCLI documentation for details.
5.2.4 DSCLIbroker maintenance

The DSCLIbroker server is an autonomic process. However, there are a few maintenance
tasks you need to perform.
The most important task is to start and stop the broker daemon using the script Broker.pl
start. After the broker daemon is started, the DSCLIbroker server is ready to receive
requests for starting sessions from any client. To clean up the DSCLIbroker server, run the
same script with the parameter stop. This parameter stops all worker processes running in
the background, and then stops the broker daemon itself.
To monitor the activities of the broker and the worker processes, a logging mechanism has
been implemented. The log files are located in the directory ~/log in the home directory of the

DSCLIbroker installation. A numbered index is appended to the log file for each day. The log
files are kept for 10 days before they are deleted.
5.2.5 Licenses
The DSCLIbroker is licensed per function and number of storage devices that are managed
by the DSCLIbroker. For this reason, the serial numbers of the storage devices must be
registered. There is no capacity-based license. For migrations, the DSCLIbroker can be
ordered as a tailored STG Lab Service, in which the DSCLIbroker is available for no additional
fee.
5.3 Programming techniques

The DSCLIbroker scripting framework provides all functions of the DS8000 Copy Services,
and some additional functions. However, the true function of DSCLIbroker is to allow you to
write your own code that takes advantage of the framework. Writing code is especially of
interest for migrations, because certain operations cannot be covered by standard tools and
software components provided with the storage devices.
5.3.1 Using DSCLIbroker scripts

The DSCLIbroker scripts are perl programs that can be used in shell scripts. These shell
scripts are the easiest way to write scripts. With this method, you can implement interactions
between applications or middle ware components and the storage devices. The following are
a few examples of what you can do with scripts:
򐂰 Stopping an application before taking a FlashCopy and then restarting the application
afterward
򐂰 Alter a database into backup mode before failing over to the remote storage system
򐂰 Implement an interactive script that guides an administrator through a migration scenario
Example 5-14 shows a simple interactive migration script.
Example 5-14 Simple interactive migration script

1 #!/bin/ksh
2 #
3
4 echo "Validating the paths"
5
6 ValidatePaths.pl -n appl_ab
7
8 echo -n "Are all paths in correct state? [y/n]"
9 read yn
10
11 if [[ $yn = 'y' ]]
12 then
13 echo "Establish Global copy to replicate"
14
15 mkpprc.pl -n appl_ab -type gcp
16 lspprc.pl -n appl_ab -waitnull
17
18 echo "All tracks has been copied"

19 echo -n "Ready to fail over to remote [y/n]"
20 read yn
21 if [[ $yn = y ]]
22 then
23 failoverpprc.pl -n appl_ab -p
24 else
25 echo "Migration aborted"
26 exit 3
27 fi
28 else
29 echo "Please fix paths issues first"
30 exit 2
31 fi
In line 6 the paths are checked for the status and if all paths defined in the repository are
available, and the results are displayed. In line 9, you are asked by the script to confirm the
paths status before the script continues. If not, the script quits with a message. Otherwise the
Global Copy is established in line 15 using the DSCLIbroker script mkpprc.pl. The
synchronization of the Global Copy is monitored in line 16 using lspprc.pl -waitnull. With
this option, the script waits until the synchronization completes. In line 19, you are asked if
everything is ready to fail over. If not the script stops with a message. If you enter yes, the
script failoverpprc.pl is applied.
You can use DSCLIbroker scripts in other script languages like python. In this case, system
calls must be used that pass the command execution to the shell.
5.3.2 Using DSCLIbroker libraries

Using DSCLIbroker libraries is the most effective way to write user customized scripts. The
libraries are written in the perl language, which means the program must also be written in
perl. No special perl programming skills are required. However, the knowledge of object and
complex data structures, and nested hashes can be helpful.
DSCLIbroker libraries provide additional functions that can be helpful to implement

automated migrations tasks. DSCLIbroker libraries can be helpful in the following situations:
򐂰 To provide the correct information so the scripts make the correct decisions and implement
a valid logic for the migration. The repository of the DSCLIbroker can contain this
information, and the libraries provide functions to retrieve it from the repository.
򐂰 To obtain current information from the storage devices, for example to retrieve the status of
paths or copy pairs.
򐂰 To run storage commands using the broker that are not related to copy services functions.
For example, you can change the host connection to a target host after a successful
replication of the data.
Retrieving information from the repository

In this example, an operation in a migration scenario requires details of the Copy Service
configuration. This information can be retrieved from the repository of the DSCLIbroker using
the library function getConfig from the module DSlib.pm (Example 5-15).
Example 5-15 Retrieve information from the repository

use DSlib;
...

my @RESULTS = getConfig ('CopyPairs', 'name', 'MycopyPairs');
for my $pairs (@RESULTS) {
$source = $pairs->{Source};
$target = $pairs->{Target};
...
...
}
In this example, the source and the target volumes of a copy relation are retrieved. Include the
DSlib library with the use statement somewhere at the top of your program. With getConfig,
the CopyPairs.cfg stanza is queried for a certain relation as shown in the example. The result
is an array of hashes, where each key of a hash element corresponds to a tag entry of the
stanza. The value in each hash entry is the data you are looking for. To unpack this data, a
simple for loop is used.
Getting information from the storage devices

Using the DSCLI, the usual way to obtain information about the storage configuration or
status information is to use an appropriate list command such as lspprc. The default output
format is a formatted table. However, when the script needs only specific values, this output
format spends extra time parsing through the data. To make processing easier, the additional
output formats stanza, delimiter, and XML are provided by the DSCLI.
The DSCLIbroker uses the XML output format from the DSCLI to parse for a required
parameter. In the following example, the Out-of-Sync values for a copy relation are collected
from the output of the lspprc.pl -sum command (Example 5-16).
Example 5-16 Parsing XML output from DSCLI

use DSPPRC;
#
# Create two new boxes because it is a remote copy
my $source=DSbox->new();
my $target=DSbox->new();
#
# Create PPRC thingy
my $PPRC=DSPPRC->new($source, $target);
#
# send 'lspprc -fmt xml' command using the pairs of 'myCopyRelation'
my @SCAN=$PPRC->scanpprc('myCopyRelation','','',-l');
#
# Collect the OOS from each pair
my $totalOOS=0;
for my $pair (@SCAN) {
totalOOS = $totalOOS + $pair->{outsynctrks};
}
print "Total number of OOS: $totalOOS\n";

The function scanpprc generates a lspprc command with the option -l for DSCLI, where the
output format XML is requested. The XML format is transformed into an array of hashes
where each hash represents the output of one pair relation from the lspprc output. To
dissolve this structure and collect the Out-of-Sync tracks, a for loop searches through all
hashes and lists the Out-of-Sync tracks as shown.
Running storage commands

The main focus of the DSCLIbroker functionality is the Copy Services functions for
FlashCopy, Remote Copy, and Global Mirror, and combinations thereof. This means that the
corresponding library modules are generating the DSCLI commands that are then sent to the
broker. The broker then forwards the commands to the worker daemon with the DSCLI in the
background.
However, the underlaying libraries where the commands are sent to the broker can be used to
issue any command. You can use them to allocate volumes, create volume groups, and so
on.
Example 5-17 shows how to set a host connection for a target host system.
Example 5-17 Making a host connection

use DSbox;
#
# Create a box object
#
my $box=DSbox->new(‘ATS_1’);
#
# Create DSCLI script
#
$box->WriteDSCLIscript('managehostconnect -volgrp v12 42');
$box->WriteDSCLIscript('lsconnect -volgrp v12 42');
#
# Execute DSCLI script
#
my $ret=$box->DoDSCLIcommand ('server');
exit($ret);
In this script, an object $box of the storage device ‘ATS_1’ is allocated. All DSCLI commands
are generated and run using this object. Any valid DSCLI command can be defined with the
function $box->WriteDSCLIscript.
Attention: All DSCLI commands that are sent to the broker must not be interactive. When
using removing commands such as rmpprc and rmpprcpath, make sure the option -quiet
is used.
In this example, two DSCLI commands are collected into a script that is stored in the object.
This DSCLI script is sent and executed to that storage device using the function
$box->DoDSCLIcommand.

5.4 Automation techniques
The DSCLIbroker provide a wide range of ways to automate migration tasks. Depending on
the type of migration and the platforms and applications involved, you can reduce an entire
migration down to a single operational task. This section covers the following automation
techniques using DSCLIbroker:
򐂰 Subjects for writing automation scripts
򐂰 Generating DSCLI scripts
򐂰 Using control files
5.4.1 Subjects for writing automation scripts

The DSCLIbroker itself offers a certain level of automation, due to the organization of the data
in the repository. The scripts included with the DSCLIbroker take advantage of this
organization, but they mostly mimic DSCLI basic commands. Using the framework of the
DSCLIbroker, automation can be more efficient, allowing a convenient and save migration
operation.
Typically a migration consists of several steps of storage operations and operations that must
be done on the servers. To determine where automation can be implemented, the technical
workflow and the responsibilities must be defined first. Typically, you have a classification of
organizational responsibilities in place, which is reflected in a dedicated storage
management, server management and application management. Typically automation does
not go across the management boundaries. A meaningful automation includes all steps that
can be handled by a single operator. End the script at the hand over to the next system
management instance in the workflow.
DSCLIbroker is typically assigned to the storage management organization. However, you

can integrate the DSCLIbroker into other storage management automation tasks. For the
storage management tasks, automation can be implemented for the following situations:
򐂰 Automating the storage migration
򐂰 Generating repository data
Automating the storage migration

The storage migration tasks are related to all tasks that must be applied against the storage
subsystems. Tasks that can be run in a row can be implemented in a script as is. If you need
to hand control over to another organization, you can have the automation script exit.
Alternately, you can have the script wait at a user prompt until control comes back to the
storage system management. The storage operator then confirms the return of the control to
the script and it can continue with the next steps.
Generally, implement simulation mode for the migration scenario so you can review the steps
of the script before they take effect. During this process, print out detailed logging information
and the DSCLI commands. In the simulation mode, no DSCLI commands are run. However,
any commands that are display information can be issued and the output checked for
accuracy before the real execution.
Generating repository data

It is mission critical for migration using the DSCLIbroker that the data in the repository is
correct. Otherwise the correct volumes for a single application or a group of applications
might not be copied to the correct target volumes. In large migration projects that take weeks
or months, the production environment will probably change due to data growth or other
reasons.

Typically, migration projects at that size are organized in waves, where groups of servers or
applications are migrated in one go. Before starting a migration of such a wave, create the
whole set of repository data for that wave. This guarantees that changes to the production
environment are considered every time.
You can maintain the repository data by implementing a script that generates the data using
the scripts mentioned in 5.1.4, “Additional useful scripts” on page 118. These scripts require a
list of volumes that are mapped to the applications and the migration wave, which is applied
next as input.
Important: You need to know which volumes are subject of the migration and which
applications are affected before generating repository data.
5.4.2 Generating DSCLI scripts

In certain cases DSCLIbroker cannot be installed in your environment because of internal
restrictions such as installation policies. In this case, you can use DSCLIbroker to support the
migration by generating the DSCLI commands for each step using simulation mode
(Example 5-12 on page 117). The output of the simulation mode of each DSCLIbroker script
can then be used as a series of steps to complete the migration.
The DSCLIbroker provides a script that can generate these DSCLI scripts automatically.
Therefore the repository data for the copy pairs, the storage devices, and optionally the paths
must be available. When all the scenarios are defined, all DSCLI commands are extracted
and collected in a control file. This control file is used to produce all DSCLI commands with
the required parameters and volumes. Example 5-18 shows that a control file containing all
commands is required for a Global Mirror migration.
Example 5-18 Control file example that created all commands for Global Mirror
#
# cmd option relation outputfile
mkpprc.pl ab LBG_mkpprc
mkflash.pl bc LBG_mkGMflash
mksession.pl ab LBG_mksession
pausepprc.pl ab LBG_pauseppc
failoverpprc.pl ab LBG_failoverpprc
reverseflash.pl -fast -tgtpprc bc LBG_reverseflash
lspprc.pl ab LBG_lspprc
lspprc.pl -r ab LBG_lspprc_remote
lspprc.pl -l ab LBG_lspprc_long
lsflash.pl bc LBG_lsflash
lsflash.pl -l bc LBG_lsflash_long
rmpprc.pl ab LBG_rmpprc
rmpprc.pl -direction reverse ab LBG_rmpprc_remote
rmsession.pl ab LBG_rmsession
The DSCLI commands are written in text files into a certain directory. Each text file holds the
commands for a single step of the scenario. Finally the DSCLI scripts themselves must be
placed in the directory where the command files are located.

5.4.3 Using control files
Using control files with the DSCLIbroker allows more automation. This automation can be
used when you are performing multiple migrations and many different scenarios apply.
Instead of writing a script or a program, control files are defined for each scenario.
In Example 5-18 on page 126, a control script was used to generate the DSCLI script. The
format of the file used in this example is generally the same as for any other control file. The
required parameters are the command itself, a list of parameters, and the stanza entry to
which the operation applies. The other fields in the control file are optional, for example a
name for a log file or a comment.
A single script can be used to read the content of a selected control file and run, step by step,
the listed commands with their parameters. The control file can be selected using a
command-line option. For each execution for a scenario, write logging information into a
dedicated log file. Every DSCLIbroker script has an option -banner, which allows you to
supply comments that are printed at the top of each execution.
The advantage of this approach is that only one program must be developed in the beginning
of the migration project. All migration scenarios are defined and maintained using the control
files. No programming is required when a scenario must be changed.
5.5 Considerations about configuration data

This addresses considerations about the data that must be provided to the DSCLIbroker. This
data is configuration data of the copy relations of the volumes to be migrated. Use the
following steps to provide the data to the DSCLIbroker:
1. Define the data input format
2. Write a script to read in the configuration data
3. Generate the repository data using the appropriate DSCLIbroker scripts
5.5.1 Data sources

For a migration using Copy Services, application data must be transmitted from a set of
source volumes to corresponding target volumes. Typically, the target volumes are on a
remote storage device. You need the volume information and the definition of the connectivity,
including the communication ports, zoning, and path relations, to extract data for the
DSCLIbroker repository
Usually you maintain this information using a data management system such as Tivoli
TotalStorage Productivity Center for Replication, or even just a spreadsheet. In any case, a
data export function that provides comma-separated value data is needed to import the data
into the DSCLIbroker repository.
The success of every migration depends on the quality of the data in the repositories. The
information in the repository must represent the exact production environment. Use
DSCLIbroker as an interface for each migration. It is a vital connectivity between the
production environment and the migration environment. Re-import the production data using
this interface into the repository of the DSCLIbroker before each migration

5.5.2 Importing data to the repository
If the migration comprises many applications and servers, the whole migration project can
take several weeks or even months. Normal production cannot be paused for that long, even
on the applications that will be migrated. Therefore, the repository data might change due to
capacity upgrades or any other configuration changes. The data in the repository of the
DSCLIbroker must be adopted accordingly before the migration can start.
Assuming you keep your configuration database at the most current state, import this data to
the DSCLIbroker repository just before the next migration is started. A script can read a csv
export of your configuration data and generate a whole new set of the DSCLIbroker repository
automatically. Use the repository generation scripts as described in “Generating repository
data” on page 125.
5.6 Example migration using DSCLIbroker

This section shows an example of a real migration that was successfully accomplished for an
important client. The example is shown according to the migration process as described in
2.3, “Data migration process” on page 18. The DSCLIbroker is engaged during the analysis
after the requirements for the needed automation are discovered.
5.6.1 Overview
The client was running several data centers in a metropolitan area. In a consolidation project,
data center sites must be migrated to a new, larger site. In this context, two DS8000 storage
devices and the connected servers must be moved to the new data center location. Both
storage devices were the primary storage of a Global Mirror configuration. The new location
was intended to be the new primary site, while the target site for the Global Mirror remains.
The whole migration comprised several thousand primary volumes and hundreds of
applications. The server platforms were mostly AIX -based, but included VMware ESX server,
OpenVMS, and zSeries server. The Global Mirror was managed using IBM TotalStorage
Productivity Center for Replication. After the migration, a new TotalStorage Productivity
Center for Replication server was established. The migration was organized in groups of
applications that must be moved one by one to the new location.

Figure 5-5 shows the basic structure of the migration.
Primary Site Secondary Site

Global Mirror
1
0
0
B
0
1
0
1
A 1
0
0 0
0
0
0
0
0 C
1
1 0
0
or
irr
lM
ba
M
ig
o
Gl
ra
te
w
to
Ne
ne
w
si
te
N
New Primary Site
Figure 5-5 Migration overview
Because of the huge number of applications, the projected duration of the whole project was
several months. The main reason for this long time was the data center logistics like change
management, hardware ordering, and installation management. Also, the client wanted the
current production to be impacted as little as possible.
The general approach was to establish another Global Copy at the new data center by
cascading the target volumes of the current Global Copy. During the initial copy phase, the
production running at the original volumes was not affected. The volumes that had to be
removed from the Global Mirror session and the direction of the new Global Copy reversed.
The Target volumes and the FlashCopy relation at the secondary site were reused for the new
Global Mirror. The new Global Mirror was started at the new data center site.
5.6.2 Analysis
In the analysis phase, the planned scenario had to be proofed for feasibility. The production
environment was analyzed and dependencies discovered. In this case, the applications had a
strong dependency on the server hardware because most applications where hosted in
logical partitions LPARS that shared hardware. Therefore, the applications had to be migrated
in groups according to their server hardware.
The original idea for the migration was to use TotalStorage Productivity Center for Replication
to establish the cascaded Global Copy relation and to transfer that relationship to the new
Global Mirror session. But it is not possible to establish the cascaded Global Copy relation to
the new data center site without removing the current Global Mirror session. Therefore all
migration steps were planned using DSCLI commands.

Given these two major topics, the analysis was that automation had to be implemented for
two major reasons:
1. To move the correct set of volumes for each application
2. To avoid human error and guarantee consistency during command execution
DSCLIbroker provides a solution for both these concerns. Mapping applications to volumes is
covered by supplying the repository with the correct data. In addition, the scripting framework
was used to write scripts that fulfill the automation requirements.
5.6.3 Test and develop

The migration environment was rebuilt in an IBM lab. This test environment was used to
develop and test the entire migration scenario. In addition to the scenarios for migration,
scenarios to roll back to normal production if problems prevent migration from completing
were.
Scripts to generate the repository data and the migration steps were developed and tested.
For the repository data, the customer provided the TotalStorage Productivity Center for
Replication session export. They also provided a spreadsheet with the applications, the
volumes, and the group they were to be migrated with. Using both data sources, the
repository data was generated for all entities.
The test environment was also used to educate the client about using the automation scripts
and provide hands-on training.
5.6.4 Planning
With the tools and the scenarios developed in the testing phase, the final planning was
finalized. All information was put in place and used to generate the instructions for every
required production change management task.
A detailed time line plan was finalized after the test and development phase. All roles were
assigned so the complete migration could be conducted by the customer.
5.6.5 Running
The migration execution was divided in to two general steps.
A trial run a couple of days before the go-live migration was scheduled. In this trial run, the
migration was run until the data was failed over from the storage perspective to the new data
center. However, production was continued at the primary site. At the new data center, the
applications were started using the migrated data and a health check was performed. After all
tests were passed, the application was ready for the go-live migration.
In the second step, the go-live migration was run. This full migration ended when the
production was started in the new data center.
5.6.6 Validating
Most validation was done during the trial run. This validation made sure that the data was
replicated as expected, which was the final approval for the go-live migration.

Even with the go-live migration, the health checks were been issued as well to double check
the success of the migration. During these validation steps, the data at the original production
was preserved in case the client wanted to back out.

6
Chapter 6. IBM XIV data migration

This chapter introduces the IBM XIV Storage System embedded data migration function,
which is used to migrate data from a non-XIV storage system to the XIV Storage System. The
XIV data migration function is included in the base XIV software, and is easy to deploy. This
chapter includes usage examples and troubleshooting information.
At a high level, the steps to migrate to XIV using the XIV Data Migration function are:
1. Establish connectivity between the source device and XIV. The source storage device
must have Fibre Channel or iSCSI connectivity with the XIV.
Important: If the IP network includes firewalls between the mirrored XIV systems, TCP
port 3260 must be open within the firewalls for iSCSI replication to work.
2. Identify in detail the configuration of the LUNs to be migrated.

3. Perform the data migration:
– Stop and unconfigure all I/O from source-original LUNs.
– Start data migration in XIV.
– Map new LUNs to the host and discover new LUNs through XIV.
– Start all I/O on the new XIV LUNs.

򐂰 Overview
򐂰 Handling I/O requests
򐂰 Data migration steps
򐂰 Command-line interface
򐂰 Manually creating the migration volume
򐂰 Changing and monitoring the progress of a migration
򐂰 Thick-to-thin migration
򐂰 Resizing the XIV volume after migration
򐂰 Troubleshooting
򐂰 Backing out of a data migration
򐂰 Migration checklist
򐂰 Other considerations

6.1 Overview
Whatever the reason for your data migration, you always want to avoid or minimize disruption
to your business applications. While there are many options for migrating data, the XIV
Storage System includes an embedded data migration feature. This feature allows you to
easily migrate data from an existing storage system to the XIV Storage System. The
production environment using the existing storage can continue functioning during the data
transfer with only one brief period of downtime. Figure 6-1 illustrates a high-level view of what
the data migration environment might look like.
Figure 6-1 Data migration simple view
The IBM XIV Data Migration solution offers a smooth data transfer for the following reasons:
򐂰 Requires only a single short outage to switch LUN ownership. This outage allows the
immediate connection of a host server to the XIV Storage System. This connection
provides you with direct access to all existing LUNs before they are copied to the XIV
Storage System.
򐂰 Synchronizes data between the two storage systems using a method not apparent to the
XIV Storage System as a background process with minimal performance impact.
򐂰 Supports data migration from practically all storage vendors.
򐂰 Can be used with Fibre Channel or iSCSI.
򐂰 Can be used to migrate SAN boot volumes.
The XIV Storage System manages the data migration by simulating host behavior. When
connected to the storage device containing the source data, XIV looks and behaves like a
SCSI initiator. In other words, it acts like a host server. After the connection is established, the
storage device containing the source data acts as though it is receiving read or write requests
from a host. In fact, the XIV Storage System is doing a block-by-block copy of the data, which
the XIV then writes onto an XIV volume.

During the background copy process, the host server is connected to the XIV Storage
System. The XIV Storage System handles all read and write requests from the host server,
even if the data is not resident on the XIV Storage System. In other words, during the data
migration, the data transfer is not apparent to the host and the data is available for immediate
access.
The connections between the two storage systems must remain intact during the entire
migration process. If at any time during the migration process the communication between the
storage systems fails, the process also fails. In addition, if communication fails after the
migration reaches synchronized status, writes from the host will fail if the source updating
option was chosen. For more information, see 6.2, “Handling I/O requests” on page 135. The
process of migrating data is performed at a volume level, as a background process.
The data migration facility in XIV firmware revisions 10.1 and later supports the following
functions:
򐂰 Up to four migration targets can be configured on an XIV. A target is either one controller in
an active/passive storage device or one active/active storage device. XIV firmware revision
10.2.2 increased the number of targets to 8. The target definitions are used for both
Remote Mirroring and data migration. Both Remote Mirroring and data migration functions
can be active at the same time. An active/passive storage device with two controllers can
use two target definitions unless only one of the controllers is used for the migration.
򐂰 The XIV can communicate with host LUN IDs ranging from 0 to 512 (in decimal). This
function does not necessarily mean that the non-XIV disk system can provide LUN IDs in
that range. You might be restricted by the non-XIV storage controller to use only 16 or 256
LUN IDs (depending on hardware vendor and device).
򐂰 Up to 4000 LUNs can be concurrently migrated.
Important: The source system is called a target when setting up paths between the XIV
Storage System and the non-XIV storage. This term is also used in Remote Mirroring, and
both functions share terminology for setting up paths for transferring data.
6.2 Handling I/O requests

The XIV Storage System handles all I/O requests for the host server during the data migration
process. Read requests are handled based on where the data is currently located. For
example, if the data is already migrated to the XIV Storage System, it is read from that
location. However, if the data is not yet migrated to the IBM XIV storage, the read request
comes from the host to the XIV Storage System. The XIV System in turn retrieves the data
from the source storage device.
6.2.1 Methods of handling write requests

The XIV Storage System handles all host server write requests, and the non-XIV disk system
is not apparent to the host. All write requests are handled using one of two user-selectable
methods, chosen when defining the data migration. The two methods are source updating
and no source updating.
Chapter 6. IBM XIV data migration 135

Selecting which method you want to use is shown in Figure 6-2. The check box must be
selected to enable source updating, shown here as Keep Source Updated. If this box is not
selected, changed data from write operations is only written to the XIV.
Figure 6-2 Keep Source Updated check box
Source updating
This method for handling write requests ensures that both storage systems are updated when
a write I/O is issued to the LUN being migrated. Source updating keeps the source system
updated during the migration process, and the two storage systems remain in sync after the
background copy process completes. The write commands are only acknowledged by the XIV
Storage System to the host after the following steps occur:
򐂰 Writing the new data to the local XIV volume
򐂰 Writing data to the source storage device
򐂰 Receiving an acknowledgement from the non-XIV storage device.
If there is a communication failure between the storage systems or a write fails, the XIV
Storage System also fails the write operation to the host. This process ensures that the
systems remain consistent. Application requirements will determine whether you use the
Keep Source Updated option.
No source updating
This method for handling write requests ensures that only the XIV volume is updated when a
write I/O is issued to the LUN being migrated. This method decreases the latency of write I/O
operations because write requests are only written to the XIV volume. This limits your ability
to back out a migration unless you have another way of recovering updates to the volume
being migrated. You can avoid this risk by shutting down the host for the duration of the
migration.
Tip: Do not select Keep Source Updated if migrating a boot LUN. Doing so allows you to
quickly back out of a migration of the boot device if a failure occurs.

6.2.2 Multi-pathing with data migrations
There are essentially two types of enterprise storage systems when it comes to multi-pathing:
򐂰 Active/active: The volumes on these storage systems can be active on all of the storage
system controllers at the same time. These systems support IO activity to any volume
down two or more paths. These types of systems typically support load balancing
capabilities between the paths with path failover and recovery in the event of a path failure.
The XIV is such a device, and can use this technology during data migrations. Examples
of IBM products that are active/active storage servers are the IBM DS6000, IBM DS8000,
IBM ESS F20, IBM ESS 800, and IBM SAN Volume Controller. DS6000 and SAN Volume
Controller are storage servers that have preferred controllers on a LUN-by-LUN basis. If
attached hosts ignore this preference, there might be a small performance penalty.
If your non-XIV disk system supports active/active, carefully configure multiple paths from
XIV to non-XIV disk. The XIV load balances the migration traffic across those paths and
automatically handles path failures.
򐂰 Active/passive: The volumes on these storage platforms can be active on only one
controller at a time. These storage devices do not support I/O activity to any volume down
multiple paths. Most support active volumes on one or more controllers at the same time,
but a volume can be active on only one controller. An example of an IBM product that is an
active/passive storage device is the IBM DS4700.
Migrating from an active/active storage device

If your non-XIV disk system supports active/active LUN access, you can configure multiple
paths from XIV to the non-XIV disk system. The XIV load balances the migration traffic across
these paths. However, do not configure more than two connections or increase the
initialization speed to a large value to speed up the migration. The XIV synchronizes one
volume at a time per target, so with four targets, four volumes can be being migrated at once.
Therefore, the speed of the migration is determined by the ability of the non-XIV storage
device to read from the LUN being migrated. Unless the non-XIV storage device has striped
the volume across multiple RAID arrays, the migration speed is unlikely to exceed 250–300
MBps. The speed is totally dependent on the non-XIV storage device.
Important: If multiple paths are created between an XIV and an active/active storage
device, the same SCSI LUN IDs must be used for each LUN on each path. Otherwise, data
corruption might occur. Configure a maximum of two paths per target because defining
more paths does not increase throughput. With some storage arrays, defining more paths
adds complexity and increases the likelihood of configuration issues and corruption.
Migrating from an active/passive storage device

Because of the active/active nature of XIV, special considerations must be made when
migrating data from an active/passive storage device to XIV. Configure a single path between
the non-XIV storage device controller and the XIV system. You might want to perform
migrations with the host applications offline due to the single path.
Define the target to the XIV per non-XIV storage controller (controller, not port). Define at
least one path from that controller to the XIV. All volumes active on the controller can be
migrated using the defined target for that controller. For example, suppose that the non- XIV
storage device contains two controllers (A and B):
򐂰 Define one target (Ctrl+A) with at least one path between the XIV and one controller on
the non-XIV storage device (controller A). All volumes active on this controller can be
migrated using this target. When defining the XIV initiator to the controller, define it as not
supporting fail-over if that option is available on the non-XIV storage array. By doing so,

volumes that are passive on the A controller are not presented to the XIV. Check your
non-XIV storage device documentation for more information.
򐂰 Define another target (Ctrl+B) with at least one path between the XIV and controller B. All
volumes active on controller B can be migrated to the XIV using this target. When defining
the XIV initiator to the controller, be sure to define it as not supporting failover if that option
is available. By doing so, volumes that are passive on controller B are not presented to the
XIV. Check your non-XIV storage device documentation for more information.
򐂰 shows the
Figure 6-3 Active/Passive as multiple targets
Tip: If your controller has two target ports (DS4700, for example), both can be defined as
links for that controller target. Make sure that the two target links are connected to separate
XIV modules. Connecting the links this way makes one redundant in a module failure.
Important: Certain examples shown in this chapter are from an IBM DS4000®
active/passive migration with each DS4000 controller defined independently as a target to
the XIV Storage System. If you define a DS4000 controller as a target, do not define the
alternative controller as a second port on the first target. Doing so causes unexpected
issues such as migration failure, preferred path errors on the DS4000, or slow migration
progress.
6.3 Data migration steps

The following are the high-level steps required when migrating a volume from a non-XIV
system to the XIV Storage System:
1. Set up the initial connection:
– Zone or cable the XIV to the non-XIV storage device.
– Define XIV to a non-XIV storage device (as a host).
– Define non-XIV storage device to XIV (as a migration device).
2. Create a data migration volume on XIV:
– Perform pre-migration tasks for the host being migrated:
• Back up your data.
• Shut down your host or application, or unmount the file system.
• Perform a point-in-time copy of original non-XIV volume if possible.
• Unzone the host from non-XIV storage.
– Define and test the data migration volume:

• On non-XIV storage, map volumes away from host and map them instead to XIV.
• On XIV, create data migration and test it.
3. Activate data migration on XIV.
4. Define the host on XIV and bring the host online:
– Zone the host to XIV.
– On XIV, map volumes to the host.
– Bring the host online:
• Update host HBA drivers and firmware.
• Install the Host Attachment Kit and detect volumes.
5. Complete the data migration:
– Monitor the migration on XIV.
– Delete the migration on XIV.
These steps are explained in detail in the following sections.
6.3.1 Initial connection setup

For the initial connection setup, zone or cable the XIV to the system being migrated.
Zoning or cabling the XIV to the non-XIV storage device

To connect the two devices, perform the following steps:
1. Run cables from port 4 on each selected XIV interface module to a fabric switch.
2. Zoned the XIV initiator ports (whose WWPNs end in 3) to the selected non-XIV storage
device host ports using single initiator zoning. Each zone contains one initiator port and
one target port.
Because the non-XIV storage device views the XIV as a host, the XIV must connect to the
non-XIV storage system as a SCSI initiator. Therefore, the physical connection from the XIV
must be from initiator ports on the XIV. The default initiator port for Fibre Channel is port 4 on
each active interface module. The initiator ports on the XIV must be fabric attached, so they
need to be zoned to the non-XIV storage system.
Use two physical connections from two separate modules on two separate fabrics for
redundancy. However, redundant pathing is not possible on active/passive controllers.
It is possible that the host might be attached through one medium (such as iSCSI), whereas
the migration occurs through the other. In this case, the host-to-XIV connection method and
the data migration connection method are independent of each other.
Depending on the non-XIV storage device vendor and device, it might be easier to zone the
XIV to the ports where the volumes being migrated are already present. Zoning in this way
might avoid needing to reconfigure the non-XIV storage device. For example, in EMC
Symmetrix/DMX environments, it is easier to zone the fiber adapters (FAs) to the XIV where
the volumes are already mapped.

Figure 6-4 depicts a fabric-attached configuration. It shows that module 4 port 4 is zoned to a
port on the non-XIV storage through Fabric A. Module 7 port 4 is zoned to a port on the
non-XIV storage through Fabric B.
Figure 6-4 Fabric attached
Defining XIV to the non-XIV storage device as a host

After the physical connection between the storage devices is complete, the XIV initiator
(WWPN) must be defined on the non-XIV storage device. This process is vendor- and
device-dependent because you must use the non-XIV storage device management interface.
See the documentation for the non-XIV storage device for information about how to configure
hosts. The XIV is seen as a host to the non-XIV storage.
If you already zoned the XIV to the non-XIV storage device, the WWPNs of the XIV initiator
ports might be displayed in the WWPN list. Whether they are displayed depends on the
non-XIV storage device and storage management software. If they are not there, you must
manually add them. The WWPNs might not be displayed if you need to map an LUN0 or if
SAN zoning is not done correctly.
The XIV must be defined as a Linux or Windows host to the non-XIV storage device. If the
non-XIV device offers several variants of Linux, you can select SuSE or RedHat Linux, or
Linux x86. Selecting the host defines the correct SCSI protocol flags for communication
between the XIV and non-XIV storage device. The principal criterion is that the host type must
start LUN numbering with LUN ID 0. If the non-XIV storage device is active/passive, check
whether the host type selected affects LUN failover between controllers. For more
information, see 6.12, “Other considerations” on page 173.
There might also be other vendor-dependent settings. For more information, see 6.12, “Other
considerations” on page 173.

Define non-XIV storage device to XIV (as a migration target)
After the physical connectivity is made and the XIV is defined to the non-XIV storage device,
the non-XIV storage device must be defined on the XIV. This process includes defining the
following aspects:
򐂰 The storage device object
򐂰 The WWPN ports on the non-XIV storage device
򐂰 The connectivity between the XIV and the non-XIV storage device.
To complete this process, perform the following steps:

1. In the XIV GUI, click Remote  Migration Connectivity.
2. Click Create Target as shown in Figure 6-5.
Tip: If Create Target is disabled, you have reached the maximum number of targets.
The number of allowed targets includes both migration and mirror targets.
Figure 6-5 Create target for the non-XIV device
3. Make the appropriate entries and selections, then click Define (Figure 6-6).
– Target Name: Enter a name of your choice.
– Target Protocol: Select FC from the list.
Figure 6-6 Defining the non-XIV storage device

4. Click the gray line to access the migration connectivity (Figure 6-7).
Tip: The data migration target is represented by an image of a generic rack. If you want
to delete or rename the migration device, right-click that rack.
Figure 6-7 Click the gray line
5. Right-click the dark box that is part of the defined target and select Add Port (Figure 6-8).
Figure 6-8 Defining the target port
6. Enter the WWPN of the first (fabric A) port on the non-XIV storage device zoned to the
XIV. There is no list, so you must manually type or paste in the correct WWPN.
Tip: You do not need to use colons to separate every second number. It makes no
difference if you enter a WWPN as 10:00:00:c9:12:34:56:78 or 100000c912345678.
7. Click Add.

8. Enter another port (repeating step 3) for those storage devices that support active/active
multi-pathing. This port can be the WWPN that is zoned to the XIV on a separate fabric.
9. Connect the XIV and non-XIV storage ports by clicking and dragging from the port on the
XIV to the port (WWPN) on the non-XIV storage device (Figure 6-9). In the example, the
mouse started at module 9 port 4 and has nearly reached the target port. The connection
turns red when the line reaches port 1 on the target.
Figure 6-9 Dragging a connection between XIV and migration target

Figure 6-10 shows the active connection from module 9 port 4 to port 1 on the non-XIV
storage device as a green line. This means that the non-XIV storage system and XIV are
connected and communicating, which includes the following concerns:
򐂰 SAN zoning is done correctly.
򐂰 The correct XIV initiator port is selected.
򐂰 The correct target WWPN is entered and selected.
򐂰 LUN 0 was detected on the target device).
If there is an issue with the path, the connection line is red.
Figure 6-10 Non-XIV storage device defined
Tip: Ensuring that LUN0 is visible on the non-XIV storage device down the controller path
that you are defining helps ensure functional connectivity. Connections from XIV to
DS4000, EMC DMX, or Hitachi HDS devices require a real disk device to be mapped as
LUN0. However, other devices, such as IBM ESS 800 and EMC CLARiiON, do not need a
LUN to be allocated to the XIV.
6.3.2 Creating a data migration volume on XIV

This section addresses the following steps needed to create a data migration volume on XIV:
򐂰 Performing pre-migration tasks for the host being migrated
򐂰 Defining and testing the data migration volume

Performing pre-migration tasks for the host being migrated
Follow these steps to prepare the host:
1. Back up the volumes being migrated.
A full restorable backup must be created before any data migration activity. Verify the
backup to make sure all the data is restorable and that there are no backup media errors.
2. Shut down the application/host.
Before the actual migration can begin, the application must be quiesced. Stopping the I/O
ensures that the application data is in a consistent state. Because the host might need to
be rebooted a number of times before the application data being available again, consider
the following steps:
– Set applications to not automatically start when the host operating system restarts.
– Stop file systems from being automatically remounted on boot. For operating systems
based on UNIX, comment out all affected file system mount points in the fstab or
vfstab.
Note: In clustered environments, you might choose to work with only one node until the
migration is complete. If so, consider shutting down all other nodes in the cluster.
3. Perform a point-in-time copy of the volume on the non-XIV storage device if that function is
available. Perform the copy before changing any host drivers or installing new host
software, particularly if you are going to migrate boot from SAN volumes.
4. Unzone the host from non-XIV storage. The host must no longer access the non-XIV
storage system after the data migration is activated. The host must perform all I/O through
the XIV.
Defining and testing the data migration volume

Perform these steps to define and test the data migration volume:
1. Allocate the non-XIV volume to XIV.
The volumes being migrated to the XIV must be allocated using LUN mapping to the XIV.
The LUN ID presented to the XIV must be a decimal value from 0 to 512. If using
hexadecimal LUN numbers, the LUN IDs can range from 0x0 to 0x200. These LUN IDs
must be converted to decimal when entered into the XIV GUI. The XIV does not recognize
a host LUN ID above 512 (decimal). Figure 6-11 shows LUN mapping using a DS4700. It
depicts the XIV as a host called XIV_Migration_Host with four DS4700 logical drives
mapped to the XIV as LUN IDs 0 - 3.
Figure 6-11 Non-XIV LUNs defined to XIV

When mapping volumes to the XIV it is important to note the LUN IDs allocated by the
non-XIV storage. The methodology to do assign LUN IDs varies by vendor and device,
and is documented in greater detail in 6.12, “Other considerations” on page 173.
Important: You must unmap the volumes from the host during this step, even if you
plan to power the host off during the migration. The non-XIV storage presents only the
migration LUNs to the XIV. Do not allow the host to detect the LUNs from both the XIV
and the non-XIV storage.
2. Define data migration object/volume.

After the volume being migrated to the XIV is allocated to the XIV, a new data migration
volume can be defined. The source volume from the non-XIV storage system and the XIV
volume must be the same size. It is easiest to allow XIV create the target LUN for you as
explained to the following section. If you want to manually create the volumes on the XIV,
see 6.5, “Manually creating the migration volume” on page 157.
Important: You cannot use the XIV data migration function to migrate data to a source
volume in an XIV remote mirror pair. If you need to use a remote mirror pair, migrate the
data first and then create the remote mirror after the migration is completed.
Creating an XIV volume automatically

The XIV can determine the size of the non-XIV volume and create the XIV volume when the
data migration object is defined. This simple method avoids potential problems when
manually calculating the real block size of a volume. To create an XIV volume automatically,
1. In the XIV GUI, click Remote  Migration.
2. Right-click Migration and select Define Data Migration.
3. In the Define Data Migration window (Figure 6-12 on page 147), make the appropriate
selections and entries:
– Destination Pool: Select the pool where the volume will be created.
– Destination Name: Enter a user-defined name that for the local XIV volume.
– Source Target System: Select the already defined non-XIV storage device.
Important: If the non-XIV device is active/passive, the source target system must
represent the controller (or service processor) on the non-XIV device that currently
owns the source LUN. Find out from the non-XIV storage which controller is
presenting the LUN to the XIV.
– Source LUN: Enter the decimal value of the host LUN ID as presented to the XIV from
the non-XIV storage system. Certain storage devices present the LUN ID as hex. The
number in this field must be the decimal equivalent. Ensure that you do not accidentally
use internal identifiers that you might also see on the source storage systems
management windows. In Figure 6-11 on page 145, the correct values are in the LUN
column.

– Keep Source Updated: Select this check box if the non-XIV storage system source
volume is to be updated with writes from the host. If you select this option, all writes
from the host are written to both volumes until the data migration object is deleted.
Note: Do not select Keep Source Updated if you are migrating the boot LUN so
that you can quickly back out if a failure occurs.
Figure 6-12 Define Data Migration object/volume
4. Click Define.
The migration begins as shown in Figure 6-13. Define Data Migration queries the
configuration of the non-XIV storage system and create an equal sized volume on XIV. To
check whether you can read from the non-XIV source volume, run Test Data Migration.
Figure 6-13 Defined data migration object/volume
Consideration: On some active/passive non-XIV storage systems, the configuration

can be read over the passive controller, but Test Data Migration fails.

5. Right-click the created data migration object and select Test Data Migration. If there are
any issues with the data migration object the test fails and the issues encountered are
reported (Figure 6-14).
Figure 6-14 Test Data Migration option
Tip: If you are migrating volumes from a Microsoft Cluster Server (MSCS) that is still
active, migration testing might fail due to reservations on the source LUN placed by MSCS.
You must bring the cluster down properly to get the test to succeed. If the cluster is not
brought down properly, errors occur either during the test or when activated. The SCSI
reservation must then be cleared for the migration to succeed.
6.3.3 Activate a data migration on XIV

After the data migration volume is tested, you can begin the actual data migration. When data
migration is initiated, the data is copied sequentially in the background from the non-XIV
storage system volume to the XIV. The host reads and writes data to the XIV storage system
without being aware of the background I/O being performed.
Important: After it is activated, the data migration can be deactivated. However, after
deactivating the data migration, the host is no longer able to read or write to the migration
volume and all host I/O stops. Do not deactivate the migration with host I/O running. If you
want to abandon the data migration, see the back-out process described in section 6.10,
“Backing out of a data migration” on page 169.
Right-click to select the data migration object/volume and choose Activate. Activate all
volumes being migrated so that they can be accessed by the host. The host has read and
write access to all volumes, but the background copy occurs serially volume by volume. If two
targets (such as non-XIV1 and non-XIV2) are defined with four volumes each, two volumes
are actively copied in the background: One volume from non-XIV1 and another from
non-XIV2. All eight volumes are accessible by the hosts.

Figure 6-15 shows the menu choices when right-clicking the data migration. The Test Data
Migration, Delete Data Migration, and Activate menu items are the most-used commands.
Figure 6-15 Activate data migration
6.3.4 Defining the host on XIV and bringing the host online
Defining the host on the XIV and bringing to online involves the following steps:
򐂰 Zoning the host to XIV
򐂰 Defining the host being migrated to the XIV
򐂰 Mapping volumes to the host on XIV
򐂰 Bringing the host online
Zoning the host to XIV

The host must first be directed using SAN fabric zoning to the XIV instead of the non-XIV
storage system. It must be directed because the XIV is acting as a proxy between the host
and the non-XIV storage system. The host must no longer access the non-XIV storage
system after the data migration is activated. The host must perform all I/O through the XIV.
Defining the host being migrated to the XIV

Before performing data migrations and allocating the volumes to the hosts, the host must be
defined on the XIV. Volumes are then mapped to the hosts or clusters. If the host is to be a
member of a cluster, the cluster must be defined first. However, a host can be easily moved
from or added to a cluster at any time. Moving the host requires that the host is zoned to your
XIV target ports through the SAN fabric.
Perform the following steps to define the host:

1. To define a cluster (optional):
a. In the XIV GUI, click Host and Clusters  Host and Clusters.
b. Click Add Cluster.
c. Enter a cluster name in the Name field.
d. Click OK.
2. To define a host:
a. In the XIV GUI, click Host and Clusters  Host and Clusters.
b. Click Add Host from the top menu bar. Make the appropriate entries and selections.
i. Name: Enter a host name.
ii. Cluster: If the host is part of a cluster, select the cluster from the list.

c. Click Add.
d. Right-click the host and select Add Port.
e. Enter the following information:
i. Port Type: Select FC from the list.
ii. Port Name: This option produces a list of WWPNs that are logged in to the XIV, but
that are not assigned to a host. WWPNs can be chosen from the list or entered
manually.
iii. Click Add.
f. Repeat these steps to add all the HBAs of the host being defined.
Mapping volumes to the host on XIV

After the data migration is started, you can use the XIV GUI or XCLI to map the migration
volumes to the host. When mapping volumes to hosts on the XIV, LUN ID 0 is reserved for
XIV in-band communication. This means that the first LUN ID that you normally use is
LUN ID 1. This restriction includes boot-from-SAN hosts. You might also choose to use the
same LUN IDs as were used on the non-XIV storage, but this is not mandatory.
Important: The host cannot read the data on the non-XIV volume until data migration is
activated. The XIV does not pass through (proxy) I/O for an inactive migration. If you use
the XCLI dm_list command to display the migrations, ensure that Yes is shown in the
Active column for every migration.
Bringing the host online

After the volumes are mapped to the host server, the host can be brought online.
The host must be configured using the XIV host attachment procedures. These procedures
include the following steps:
򐂰 Removing any existing/non-XIV multi-pathing software
򐂰 Installing the native multi-pathing drivers and recommended patches,
򐂰 Installing the XIV Host attachment kit, as identified in the XIV Host Attachment Guides
򐂰 Installing the most current HBA driver and firmware
One or more reboots might be required. Documentation and software can be found at:
http://www.ibm.com/support/search.wss?q=ssg1*&tc=STJTAG+HW3E0&rs=1319&dc=D400&dtm
When volume visibility is verified, the application can be brought up and operations verified.
Tip: In clustered environments, bring only one node of the cluster online initially after the
migration is started. Leave all other nodes offline until the migration is complete. After the
migration is complete, update all other nodes (driver, host attachment package, and so on),
in the same way as the primary node. For more information, see “Performing pre-migration
tasks for the host being migrated” on page 145.
6.3.5 Completing the data migration on XIV

To complete the data migration, perform the following steps:
򐂰 Tracking the data migration progress
򐂰 Deleting data migration

Tracking the data migration progress
Figure 6-16 shows the progress of the data migrations. The status bar can be toggled
between GB remaining, percent complete, and hours/minutes remaining. The example shows
four data migrations, one of which has started background copy and three of which have not.
Only one migration is being copied at this same time because there is only one target
(DS4700_Ctrl_B).
Figure 6-16 Data migration progress
After all of the data in a volume is copied, the data migration achieves synchronization status.
After synchronization is achieved, all read requests are served by the XIV Storage System. If
source updating was selected, the XIV continues to write data to both itself and the outgoing
storage system until the data migration is deleted. Figure 6-17 shows a completed migration.
Figure 6-17 Data migration complete
Deleting data migration

After synchronization is achieved, the data migration object can be safely deleted without host
interruption.
Important: If performing an online migration, do not deactivate the data migration before
deletion. Deactivating before deletion causes host I/O to stop, which can cause data
corruption.

Right-click the data migration volume and select Delete Data Migration as shown in
Figure 6-18. Deleting the migration can be done without host/server interruption.
Figure 6-18 Delete Data Migration option
Restriction: For safety purposes, you cannot delete an inactive or unsynchronized data
migration from the Data Migration panel. An unfinished data migration can be deleted only
by deleting the relevant volume from the Volumes  Volumes & Snapshots section in the
XIV GUI.
6.4 Command-line interface

All of the XIV GUI operation steps can be performed using the XIV command-line interface
(XCLI). You can use XCLI either through direct command execution or through batch files
containing multiple commands. Using the CLI is especially helpful in migration scenarios
involving numerous LUNs. This section lists the XCLI commands equivalent to the GUI steps
shown previously. A full description of all the XCLI commands can be found in the XCLI Users
Guide available at:
http://publib.boulder.ibm.com/infocenter/ibmxiv/r2/topic/com.ibm.help.xiv.doc/docs
/GC27-2213-02.pdf
When using the XIV GUI, every command issued is logged in a text file with the correct syntax
so you can review it later. This log is helpful for creating scripts. If you are running the XIV GUI
under Microsoft Windows, look for a file named guicommands_< todays date >.txt. This file is
typically found in the following folder:
C:\Documents and Settings\ < Windows user ID >\Application Data\XIV\GUI10\logs
The commands in the next few pages are in the order in which you must run them, starting
with the commands to list all current definitions. You need these definitions when you delete
migrations.
򐂰 List targets.
Syntax target_list
򐂰 List target ports.
Syntax target_port_list

򐂰 List target connectivity.
Syntax target_connectivity_list
򐂰 List clusters.
Syntax cluster_list
򐂰 List hosts.
Syntax host_list
򐂰 List volumes.
Syntax vol_list
򐂰 List data migrations.
Syntax dm_list
򐂰 Define target (Fibre Channel only).
Syntax target_define target=<Name> protocol=FC xiv_features=no
Example target_define target=DMX605 protocol=FC xiv_features=no
򐂰 Define target port (Fibre Channel only).
Syntax target_port_add fcaddress=<non-XIV storage WWPN>
target=<Name>
Example target_port_add fcaddress=0123456789012345 target=DMX605
򐂰 Define target connectivity (Fibre Channel only).
Syntax target_connectivity_define
local_port=1:FC_Port:<Module:Port> fcaddress=<non-XIV
storage WWPN> target=<Name>
Example target_connectivity_define local_port=1:FC_Port:5:4
fcaddress=0123456789012345 target=DMX605
򐂰 Define cluster (optional).
Syntax cluster_create cluster=<Name>
Example cluster_create cluster=Exch01
򐂰 Define host (if adding host to a cluster).
Syntax host_define host=<Host Name> cluster=<Cluster Name>
Example host_define host=Exch01N1 cluster=Exch01
򐂰 Define host (if not using cluster definition).
Syntax host_define host=<Name>
Example host_define host=Exch01
򐂰 Define host port (Fibre Channel host bus adapter port).
Syntax host_add_port host=<Host Name> fcaddress=<HBA WWPN>
Example host_add_port host=Exch01 fcaddress=123456789abcdef1
򐂰 Create XIV volume using decimal GB volume size.
Syntax vol_create vol=<Vol name> size=<Size> pool=<Pool Name>
Example vol_create vol=Exch01_sg01_db size=17 pool=Exchange

򐂰 Create XIV volume using 512-byte blocks.
Syntax vol_create vol=<Vol name> size_blocks=<Size in blocks>
pool=<Pool Name>
Example vol_create vol=Exch01_sg01_db size_blocks=32768
pool=Exchange
򐂰 Define data migration.
If you want the local volume to be automatically created:
Syntax dm_define target=<Target> vol=<Volume Name> lun=<Host LUN
ID as presented to XIV> source_updating=<yes|no>
create_vol=yes pool=<XIV Pool Name>
Example dm_define target=DMX605 vol=Exch01_sg01_db lun=5
source_updating=no create_vol=yes pool=Exchange
If the local volume was pre-created:
Syntax dm_define target=<Target> vol=<Pre-created Volume Name>
lun=<Host LUN ID as presented to XIV>
source_updating=<yes|no>
Example dm_define target=DMX605 vol=Exch01_sg01_db lun=5
source_updating=no
򐂰 Test data migration object.
Syntax dm_test vol=<DM Name>
Example dm_test vol=Exch_sg01_db
򐂰 Activate data migration object.
Syntax dm_activate vol=<DM Name>
Example dm_activate vol=Exch_sg01_db
򐂰 Map volume to host/cluster.
– Map to host:
Syntax map_vol host=<Host Name> vol=<Vol Name> lun=<LUN ID>
Example map_vol host=Exch01 vol=Exch01_sg01_db lun=1
– Map to cluster:
Syntax map_vol host=<Cluster Name> vol=<Vol Name> lun=<LUN ID>
Example map_vol host=Exch01 vol=Exch01_sg01_db lun=1
򐂰 Delete data migration object.
If the data migration is synchronized and thus completed:
Syntax dm_delete vol=<DM Volume name>
Example dm_delete vol=Exch01_sg01_db
If the data migration is not complete, delete it by removing the corresponding volume from
the Volume and Snapshot menu or using the vol_delete command.
򐂰 Delete volume (not normally needed).
This command must be acknowledged by the user before running. It therefore cannot be
done using a script:
Syntax vol_delete vol=<Vol Name>
Example vol_delete vol=Exch_sg01_db

If you want to perform an unchallenged volume deletion:
Syntax vol_delete -y vol=<Vol Name>
Example vol_delete -y vol=Exch_sg01_db
򐂰 Delete target connectivity.
Syntax target_connectivity_delete
local_port=1:FC_Port:<Module:Port> fcaddress=<non-XIV
storage device WWPN> target=<Name>
Example target_connectivity_delete local_port=1:FC_Port:5:4
fcaddress=0123456789012345 target=DMX605
򐂰 Delete target port.
Fibre Channel
Syntax target_port_delete fcaddress=<non-XIV WWPN> target=<Name>
Example target_port_delete fcaddress=0123456789012345 target=DMX605
򐂰 Delete target.
Syntax target_delete target=<Target Name>
Example target_delete target=DMX605
򐂰 Change Migration Sync Rate
Syntax target_config_sync_rates target=<Target Name>
max_initialization_rate=<Rate in MB>
Example target_config_sync_rates target=DMX605
max_initialization_rate=100
6.4.1 Using XCLI scripts or batch files

To run an XCLI batch job, use the XCLI rather than a background XCLI Session.
Setting environment variables in Windows

You can remove the need to specify user and password information for every command by
making that information an environment variable at the Windows OS level. Example 6-1
shows how to set variables using a Windows command prompt. In the example, the
XIV_XCLIUSER variable is set to admin, and the XIV_XCLIPASSWORD to adminadmin.
Both variables are then confirmed as set. If necessary, change the user ID and password to
suit your setup.
Example 6-1 Setting environment variables in Microsoft Windows

C:\>set XIV_XCLIUSER=admin
C:\>set XIV_XCLIPASSWORD=adminadmin
C:\>set | find "XIV"
XIV_XCLIPASSWORD=adminadmin
XIV_XCLIUSER=admin
To make these changes permanent, perform the following steps:

1. Right-click the My Computer icon and select Properties.
2. Click the Advanced tab.
3. Click Environment Variables.
4. Click New for a new system variable.
5. Create the XIV_XCLIUSER variable with the relevant user name.

6. Click New again to create the XIV_XCLIPASSWORD variable with the relevant password.
Setting environment variables in UNIX

If you are using an operating system based on UNIX, export the environment variables as
shown in Example 6-2, which in this example is AIX. The user and password variables are set
to admin and adminadmin, and then confirmed as being set.
Example 6-2 Setting environment variables in UNIX

root@dolly:/tmp/XIVGUI# export XIV_XCLIUSER=admin
root@dolly:/tmp/XIVGUI# export XIV_XCLIPASSWORD=adminadmin
root@dolly:/tmp/XIVGUI# env | grep XIV
XIV_XCLIPASSWORD=adminadmin
XIV_XCLIUSER=admin
To make these changes permanent, update the relevant profile, making sure that you export
the variables to make them environment variables.
Tip: It is also possible to run XCLI commands without setting environment variables with
the -u and -p switches.
6.4.2 Sample scripts

With the environment variables set, a script or batch file like the one in Example 6-3 can be
run from the shell or command prompt. The script defines the data migration pairings.
Example 6-3 Data migration definition batch file

xcli -m 10.10.0.10 dm_define vol=MigVol_1 target=DS4200_CTRL_A lun=4
source_updating=no create_vol=yes pool=test_pool
Run an equivalent script or batch job to run the data migrations as shown in Example 6-4.
Example 6-4 Activate data migration batch file

xcli -m 10.10.0.10 dm_activate vol=MigVol_1

6.5 Manually creating the migration volume

The local XIV volume can be created before defining the data migration object. However, this
method is prone to manual calculation errors. These errors occur because it requires the size
of the source volume on the non-XIV storage device in 512-byte blocks. The two volumes
(source and XIV volume) must be exactly the same size. Finding the actual size of a volume
in blocks or bytes can be difficult because certain storage devices do not show the exact
volume size. You can rely on the host operating system to provide the real volume size, but
this size is also not always reliable.
As an example, consider the ESS 800 volume 00F-FCA33 as depicted in Figure 6-26 on
page 167. The size reported by the ESS 800 web GUI is 10 GB, which suggests that the
volume is 10,000,000,000 bytes in size. The ESS 800 displays volume sizes using decimal
counting. The AIX bootinfo -s hdisk2 command reports the volume as 9,536 GiB, which is
9,999,220,736 bytes (1,073,741,824 bytes per GiB). Both of these values are too small.
When the volume properties are viewed on the volume information window of the ESS 800
Copy Services GUI, it correctly reports the volume as being 19,531,264 sectors. This number
is equivalent to 10,000,007,168 bytes (512 bytes per sector). When the XIV automatically
creates a volume to migrate the contents of 00F-FCA33, it creates it as 19,531,264 blocks. Of
the three information sources considered to calculate volume size, only one of them is
correct. Using the automatic volume creation eliminates this uncertainty.
After you determine the exact size, select Blocks from the Volume Size list and enter the
size of the XIV volume in blocks. If your sizing calculation is correct, you create an XIV volume
that is the same size as the source (non-XIV storage device) volume. Then you can define a
migration:
1. In the XIV GUI, click Remote  Migration.
2. Right-click Migration and select Define Data Migration. Make the appropriate entries
and selections:
– Destination Pool: Select the pool where the volume was created.
– Destination Name: Select the pre-created volume from the list.
– Source Target System: Select the already defined non-XIV storage device from the
list.
Important: If the non-XIV device is active/passive, the source target system must
represent the controller (or service processor) on the device that owns the source
LUN. You must find out from the non-XIV storage which controller is presenting the
LUN to the XIV.

– Source LUN: Enter the decimal value of the LUN as presented to the XIV from the
non-XIV storage system. Certain storage devices present the LUN ID as hex. The
number in this field must be the decimal equivalent.
– Keep Source Updated: Select this check box if the non-XIV storage system source
volume is to be updated with writes from the host. All writes from the host are written to
the XIV volume and the non-XIV source volume until the data migration object is
deleted.
3. Click Define.
4. Test the data migration object. Right-click the created data migration volume and select
Test Data Migration. If there are any issues with the data migration object, the test fails
and reports the issue found.
If the volume that you created is the wrong size, an error message is issued during the test
data migration as shown in Figure 6-19. If you activate the migration, you get the same error
message. You must delete the volume on the XIV and create a new, correctly sized one. It
must be deleted because you cannot resize a volume that is in a data migration pair. In
addition, you cannot delete a data migration pair unless it has completed the background
copy. Delete the volume and then investigate why your size calculation was wrong. After
correcting the problem, create a new volume and a new migration, and test it again.
Figure 6-19 XIV volume wrong size for migration
6.6 Changing and monitoring the progress of a migration

You can adjust the speed of the migration process to improve performance or reduce the
impact of the migration on the XIV subsystem itself. You can also monitor its rate. Take care
when planning the migration to avoid affecting the performance of the overall system and
production I/O.

6.6.1 Changing the synchronization rate
There is only one tunable parameter that determines the speed at which migration data is
transferred between the XIV and defined targets. There are two other tunable parameters that
apply to XIV Remote Mirroring:
򐂰 max_initialization_rate
This parameter sets the rate (in MBps) at which data is transferred between the XIV and
the defined targets. The default rate is 100 MBps, and can be configured on a per-target
basis. In this example, a total of 150 MBps (100+50) transfer rate is possible. If the transfer
rate is lower than the initialization rate, You might be exceeding the capabilities of the
non-XIV disk system. If the migration is not being done with attached hosts offline,
consider dropping the initialization rate to a low number. Adjusting the rate ensures that
the volume of migration I/O does not interfere with other hosts using the non-XIV disk
system. You can then slowly increase the number while checking to ensure that response
times are not affected on the other attached hosts. If you set the max_initialization_rate to
zero, you stop the background copy, but hosts are still able to access all activated
migration volumes.
򐂰 max_syncjob_rate
This parameter (in MBps) is used in XIV remote mirroring for synchronizing mirrored
snapshots. It is not normally relevant to data migrations. However, the
max_initialization_rate cannot be greater than the max_syncjob_rate, which in turn cannot
be greater than the max_resync_rate. In general, do not increase this rate.
򐂰 max_resync_rate
This parameter (in MBps) is used for XIV remote mirroring only. It is not normally relevant
to data migrations. This parameter defines the resync rate for mirrored pairs. After
remotely mirrored volumes are synchronized, a resync is required if the replication is
stopped for any reason. This parameter affects only this resync. The default rate is 300
MBps. There is no minimum or maximum rate. However, setting the value to 400 or more
in a 4-Gbps environment does not show any increase in throughput. In general, do not
increase this rate.
Increasing the max_initialization_rate parameter might decrease the time required to migrate
the data. However, doing so might affect existing production servers on the non-XIV storage
device. By increasing the rate parameters, more outgoing disk resources are used to serve
migrations and less for existing production I/O. Be aware of how these parameters affect
migrations and production. You can choose to use the higher rate only during off-peak
production periods.
The rate parameters can only be set using XCLI, not the XIV GUI. Run the target_list -x
command, where the -x parameter displays the current rate. If the setting is changed, the
change takes place dynamically, so you do not need to deactivate or activate the migrations.
As shown in Example 6-5, first display the target list and then confirm the current rates using
the -x parameter. The example shows that the initialization rate is still set to the default value
(100 MBps). You then increase the initialization rate to 200 MBps. You can then observe the
completion rate, as shown in Figure 6-16 on page 151, to see whether it has improved.
Example 6-5 Displaying and changing the maximum initialization rate

>> target_list
Name SCSI Type Connected
Nextrazap ITSO ESS800 FC yes
>> target_list -x target="Nextrazap ITSO ESS800"
<XCLIRETURN STATUS="SUCCESS" COMMAND_LINE="target_list -x target="Nextrazap
ITSO ESS800"">

<OUTPUT>
<target id="4502445">
<id value="4502445"/>
<creator value="xiv_maintenance"/>
<creator_category value="xiv_maintenance"/>
<name value="Nextrazap ITSO ESS800"/>
<scsi_type value="FC"/>
<xiv_target value="no"/>
<iscsi_name value=""/>
<connected value="yes"/>
<port_list value="5005076300C90C21,5005076300CF0C21"/>
<num_ports value="2"/>
<system_id value="0"/>
<max_initialization_rate value="100"/>
<max_resync_rate value="300"/>
<max_syncjob_rate value="300"/>
<connectivity_lost_event_threshold value="30"/>
<xscsi value="no"/>
</target>
</OUTPUT>
</XCLIRETURN>
>> target_config_sync_rates target="Nextrazap ITSO ESS800"
max_initialization_rate=200
Command executed successfully.
Important: Increasing the initialization rate does not necessarily increase the actual speed
of the copy. The outgoing disk system or the SAN fabric might be the limiting factor. In
addition, you might cause host system impact by committing too much bandwidth to
migration I/O.
6.6.2 Monitoring migration speed

If you want to monitor the speed of the migration you can use the Data Migration window as
shown in Figure 6-16 on page 151. The status bar can be toggled between GB remaining,
percent complete, or hours/minutes remaining. However, if you want to monitor the actual
MBps, you must use an external tool. An external tool is needed is because the performance
statistics displayed using the XIV GUI or using an XIV tool do not include data migration I/O.
They do, however, show incoming I/O rates from hosts using LUNs that are being migrated.
6.6.3 Monitoring migration using the XIV event log

The XIV event log can be used to confirm when a migration started and finished using the
following steps:
1. From the XIV GUI, click Monitor  Events.
2. On the Events window, select dm in the Type list and click Filter.

Figure 6-20 shows the events for a single migration. In this example, the events must be read
from bottom to top. You can sort the events by date and time by clicking the Date column in
the Events window.
Figure 6-20 XIV Event GUI
6.6.4 Monitoring migration speed through the fabric

If you have a Brocade-based SAN, use the portperfshow command and verify the throughput
rate of the initiator ports on the XIV. If you have two fabrics, you might need to connect to two
different switches. If multiple paths are defined between XIV and non-XIV disk system, the
XIV load balances across those ports. This balancing means that you must aggregate the
throughput numbers from each initiator port to see total throughput. Example 6-6 shows the
output of the portperfshow command. The values shown are the combined send and receive
throughput in MBps for each port. In this example, port 0 is the XIV Initiator port and port 1 is
a DS4800 host port. The max_initialization_rate is set to 50 MBps.
Example 6-6 Brocade portperfshow command

FB1_RC6_PDC:admin> portperfshow
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Total
======================================================================================
50m 50m 14m 14m 2.4m 848k 108k 34k 0 937k 0 27m 3.0m 0 949k 3.0m 125m
If you have a Cisco-based SAN, start Device Manager for the relevant switch and then select
Interface  Monitor  FC Enabled.
6.6.5 Monitoring migration speed through the non-XIV storage

The ability to display migration throughput varies by non-XIV storage device. For example, if
you are migrating from a DS4000, you can use the performance monitoring windows in the
DS4000 System Manager. In the DS4000 System Manager GUI, click Storage
Subsystem  Monitor Performance to show the volumes being migrated and the
throughput for the relevant controllers. You can then determine what percentage of I/O is

being generated by the migration process. In Figure 6-21, you can see that one volume is
being migrated using a max_initialization_rate of 50 MBps. This volume represents the bulk of
the I/O being serviced by the DS4000 in this example.
Figure 6-21 Monitoring a DS4000 migration
6.7 Thick-to-thin migration

When the XIV migrates data from a LUN on a non-XIV disk system to an XIV volume, it reads
every block of the source LUN. However, when it comes to writing this data into the XIV
volume, the XIV writes only blocks that contain data. Blocks that contain only zeros are not
written and do not take any space on the XIV. This process is called a thick-to-thin migration,
and occurs regardless of whether you are migrating the data into a thin provisioning pool or a
regular pool.
During the migration, the value displayed in the Used column of the Volumes and Snapshots
window drops every time empty blocks are detected. When the migration is completed, you
can check this column to determine how much real data was written into the XIV volume. In
Figure 6-22, the used space on the Windows2003_D volume is 4 GB. However, the Windows
file system using this disk shown in Figure 6-24 on page 164 shows only 1.4 GB of data.
Figure 6-22 Thick-to-thin results
This discrepancy occurs because when file deletions occur at a file system level, the data is
not removed. The file system reuses this effectively free space. However, the system does not
write zeros over the old data because doing so generates a large amount of unnecessary I/O.
The result is that the XIV copies old and deleted data during the migration. Copying this
obsolete data makes no difference to the migration speed because the blocks must be read
into the cache regardless of what they contain.
If you are not planning to use the thin provisioning capability of the XIV, this is not an issue.
6.7.1 Writing zeros to recover space

One way to recover space before you start a migration is to use a utility to write zeros across
all free space. In a UNIX environment, use a simple script like the one shown in Example 6-7
to write large empty files across your file system. You might need to run these commands
many times to use up all the empty space.
Example 6-7 Writing zeros across your file system

# The next command will write a 1 GB mytestfile.out
dd if=/dev/zero of=mytestfile.out bs=1000 count=1000000

# The next command will free the file allocation space
rm mytestfile.out
In a Windows environment, you can use a Microsoft tool known as sdelete to write zeros
across deleted files. You can find this tool in the sysinternals section of Microsoft Technet at
the following web address:
http://technet.microsoft.com/en-us/sysinternals/bb897443.aspx
6.7.2 Recovering space after the migration

If you instead choose to write zeros to recover space after the migration, you must initially
generate large amounts of empty files. It takes several days for the used space value to
decrease after the script or application is run. It takes extra time for because recovery of
empty space runs as a background task.
6.8 Resizing the XIV volume after migration

Because of the way that XIV distributes data, the XIV allocates space in 17-GB portions
(which are exactly 17,179,869,184 bytes or 16 GiB). When creating volumes using the XIV
GUI, any volume size you enter gets rounded up to the next 17-GB cutoff.
If you to allow the XIV to determine the size of the migration volume, a small amount of extra
space is consumed for every volume that was created. The volumes automatically created by
the XIV often reserve more XIV disk space than is made available to the volume. Avoid this
problem by creating volume sizes on the non-XIV storage device in multiples of 16 GiB. An
example of the XIV volume properties of such an automatically created volume is shown in
Figure 6-23. In this example, the Windows2003_D drive is 53 GB in size, but the size on disk
is 68 GB on the XIV.
Figure 6-23 Properties of a migrated volume

This volume can be resized to 68 GB, which makes the volume 15 GB larger without
consuming any more space on the XIV. Figure 6-24 shows that the migrated Windows2003_D
drive is 53 GB in size (53,678,141,440 bytes).
Figure 6-24 Windows D drive at 53 GB
To resize a volume, perform the following steps:

1. Click Volumes  Volumes & Snapshots.
2. Right-click to select the volume and select Resize.
3. Change the sizing method from Blocks to GB.
The volume size is automatically moved to the next multiple of 17 GB.
You can also use XCLI commands as shown in Example 6-8.
Example 6-8 Resize the D drive using XCLI

>> vol_resize vol=Windows2003_D size=68
Warning: ARE_YOU_SURE_YOU_WANT_TO_ENLARGE_VOLUME Y/N: Y

Command executed successfully.
Because this example is for a Microsoft Windows 2003 basic NTFS disk, you can use the
diskpart utility to extend the volume (Example 6-9).
Example 6-9 Expanding a Windows volume

C:\>diskpart
DISKPART> list volume
Volume ### Ltr Label Fs Type Size Status Info

---------- --- ----------- ----- ---------- ------- --------- --------
Volume 0 C NTFS Partition 34 GB Healthy System
Volume 4 D Windows2003 NTFS Partition 64 GB Healthy
DISKPART> select volume 4

Volume 4 is the selected volume.
DISKPART> extend
DiskPart successfully extended the volume
Confirm that the volume has grown by displaying the volume properties. In Figure 6-25, you
can see that the disk is now 68 GB (68,713,955,328 bytes).
Figure 6-25 Windows 2003 D drive has grown to 64 GB
In terms of when to do the resize, a volume cannot be resized while it is part of a data
migration. The migration process must complete and the migration for that volume must be
deleted before the volume can be resized. For this reason, you might choose to defer resizing
until after the migration of all relevant volumes is complete. This technique also separates the
resize change from the migration change. Depending on the operating system using that
volume, you might not get any benefit from doing this resize.
6.9 Troubleshooting
This section lists common errors that are encountered during data migrations using the XIV
data migration facility.
6.9.1 Target connectivity fails

The connections (link line) between the XIV and non-XIV disks system on the migration
connectivity panel remain colored red or the link shows as down. There are several reasons
that the connection can fail:
򐂰 On the Migration Connectivity window, verify that the status of the XIV initiator port is OK
(Online). If not, check the connections between the XIV and the SAN switch.
򐂰 Verify that the Fibre Channel ports on the non-XIV storage device are set to target,
enabled, and online.
򐂰 Check whether SAN zoning is incorrect or incomplete. Verify that SAN fabric zoning
configuration for XIV and non-XIV storage device are active.

򐂰 Check the SAN switch name server that both XIV ports and non-XIV storage ports are
logged in correctly. Verify that XIV and non-XIV are logged in to the switch at the correct
speed.
򐂰 Check whether the XIV WWPN is not properly defined to the non-XIV storage device
target port. The XIV WWPN must be defined as a Linux or Windows host:
– If the XIV initiator port is defined as a Linux host to the non-XIV storage device, change
the definition to a Windows host. Delete the link (line connections) between the XIV
and non-XIV storage device ports and redefine the link. This step is storage device
dependent. The difference is caused by how the non-XIV storage device presents a
pseudo LUN-0 if a real volume is not presented as LUN 0.
– If the XIV initiator port is defined as a Windows host to the non-XIV storage device,
change the definition to a Linux host. Delete the link (line connections) between the
XIV and non-XIV storage device ports and redefine the link. This step is storage device
dependent. The difference is caused by how the non-XIV storage device presents a
pseudo LUN-0 if a real volume is not presented as LUN 0.
– If the previous two attempts are not successful, assign a real disk/volume to LUN 0 and
present to the XIV. The volume assigned to LUN-0 can be a small unused volume or a
real volume that will be migrated.
򐂰 Offline/Online the XIV Fibre Channel port:
a. Go to the Migration Connectivity window.
b. Click the link between XIV and the target system.
c. Right-click the port in question and select Configure.
d. Select No in the Enabled list.
e. Click Configure.
f. Repeat the process, selecting Yes for Enabled.
򐂰 Change the port type from Initiator to Target and then back to Initiator. This forces the port
to completely reset and reload.
a. Go to the Migration Connectivity window.
b. Clicking the link between the XIV and target system.
c. Right-click the port in question and select Configure.
d. Select Target in the Role list.
e. Click Configure.
f. Repeat the process, selecting Initiator for the role.
6.9.2 Remote volume LUN is unavailable

This error typically occurs when defining a data migration volume and the LUN ID specified in
the Source LUN field is not responding to the XIV. The error can occur for several reasons:
򐂰 The LUN ID (host LUN ID or SCSI ID) is not allocated to the XIV on the ports identified in
the target definition. Log on to the non-XIV storage device to confirm.
򐂰 The LUN ID is not allocated to the XIV on all ports specified in the target definition. For
example, if the target definition has two links, the volume must be allocated down both
paths using the same LUN ID. The XIV looks for the LUN ID specified on the first defined
path. If it does not have access to the LUN, it fails even if the LUN is allocated down the
second path. If two links are defined from the target (non-XIV) storage device to the XIV,
the LUN must be allocated down both paths.
򐂰 The LUN ID is incorrect. Do not confuse the internal LUN ID of a non-XIV storage device
with the SCSI LUN ID (host LUN ID) that is presented to the XIV. The source LUN must be
the LUN ID (decimal) as presented to the XIV.

򐂰 The Source LUN ID field is expecting a decimal number. Certain vendors present the LUN
ID in hex, must be translated to decimal. Therefore, if LUN ID 10 displays its IDs in hex,
the LUN ID in the data migration define is 16 (hex 10). An example of a hexadecimal LUN
number is shown in Figure 6-26 on page 167. In this example you can see LUN 000E,
000F, and 0010 on an ESS 800. These LUNs are entered into the XIV data migration
definitions as LUNs 14, 15, and 16. See 6.12, “Other considerations” on page 173, for
more details.
򐂰 The LUN ID allocated to the XIV is allocated to an incorrect XIV WWPN. Make sure that
the volume is allocated to the correct XIV WWPNs.
򐂰 If multiple data migration targets are defined, the wrong target might have been chosen
when the data migration was defined.
򐂰 Sometimes when volumes are added after the initial connectivity is defined, the volume is
not available. Go to the Migration Connectivity window and delete the links between the
XIV and non-XIV storage device and recreate them. Go back to the DM window and
recreate the data migration. For more information, see “Define non-XIV storage device to
XIV (as a migration target)” on page 141.
Figure 6-26 ESS 800 LUN numbers
򐂰 The volume on the source non-XIV storage device might not be initialized or been
low-level formatted. If the volume has data on it, this is not the case. However, if you are
assigning new volumes from the non-XIV storage device, these new volumes might not
complete the initialization process. On ESS 800 storage, the initialization process can be
displayed from the Modify Volume Assignments window. Figure 6-26 shows the volumes
are still 0% background formatted, so they are not accessible by the XIV. For ESS 800,
keep clicking Refresh Status on the ESS 800 web GUI until the formatting message
disappears.
6.9.3 Local volume is not formatted

This error occurs when an existing volume is chosen as the destination name that is already
written either from a host or a previous DM process that was removed. To get around this
error, perform one of the following actions:
򐂰 Use another volume as a migration destination.
򐂰 Delete the volume that you are trying to migrate to and then create it again.
򐂰 Click Volumes  Volumes and Snapshots, right-click the volume and select Format.

Important: This action deletes all data currently on the volume without recovery. You
will see a warning message that challenges the request.
6.9.4 Host server cannot access the XIV migration volume

This error occurs if you read the contents of a volume on a non-XIV storage device using an
XIV data migration without activating the data migration. It also happens if the migration is
performed without following the correct order of steps. Do not attempt to access the XIV
volume being migrated until the XIV shows that the migration is initializing and active, or fully
synchronized.
Tip: This error might also happen in a cluster environment where the XIV is holding a SCSI
reservation. Make sure that all nodes of a cluster are shut down before starting a migration.
The XCLI command reservation_list lists all SCSI reservations held by the XIV. If you
find a volume with reservations where all nodes are offline, remove the reservations using
the XCLI command reservation_clear. See the XCLI documentation for further details.
6.9.5 Remote volume cannot be read

This error occurs when a volume is defined down the passive path on an active/passive
multi-pathing storage device. It can occur in several cases:
򐂰 Two paths were defined on a target (non-XIV storage device) that supports only
active/passive multi-pathing. XIV is an active/active storage device. Defining two paths on
any target from an active/passive multi-pathing storage device is not supported.
Redefine the target with only one path. Another target can be defined with one connection
to the other controller. For example, consider an environment where the non-XIV storage
device has two controllers, but the volume can be active on only one at time. In this
environment, controller A can be defined as one target on the XIV and controller B can be
defined as a separate target. In this manner, all active volumes on controller A can be
migrated down the XIV A target. Similarly, all volumes active on the B controller can be
migrated down the XIV B target.
򐂰 When defining the XIV initiator to an active/passive multi-pathing non-XIV storage device,
certain storage devices allow the initiator to be defined as not supporting failover.
Configure the XIV initiator to the non-XIV storage device in this manner. When configured
as such, the volume on the passive controller is not presented to the initiator (XIV). The
volume is only presented down the active controller.
For more information, see 6.2.2, “Multi-pathing with data migrations” on page 137, and 6.12,
“Other considerations” on page 173.
6.9.6 LUN is out of range

XIV currently supports migrating data from LUNs with a LUN ID less than 513 (decimal). This
limitation is not usually an issue because most non-XIV storage devices present volumes on
an initiator basis. If three hosts are connected to the same port on a non-XIV storage device,
each host can be allocated volumes starting at the same LUN ID. For migration purposes, you
must do one of the following procedures:
򐂰 Map one host at a time (and then reuse the LUN IDs for the next host)
򐂰 Use different sequential LUN numbers

If three hosts each have three LUNs mapped using LUN IDs 20, 21, and 22, for migration
purposes, migrate them using the following LUN IDs:
򐂰 IDs 30, 31, 32 (first host)
򐂰 IDs 33, 34, 35 (second host)
򐂰 IDs 36, 37, 38 (third host)
From the XIV you can again map them to each host as LUN IDs 20, 21, and 22 (as they were
from the non-XIV storage).
If migrating from an EMC Symmetrix or DMX, there are special considerations. See 6.12,
“Other considerations” on page 173.
6.10 Backing out of a data migration

For change management purposes, you might be required to document a back-out
procedure. There are four possible points in the migration process where a back-out might
occur:
򐂰 Back out before migration is defined on the XIV
򐂰 Back out after a data migration is defined but not activated
򐂰 Back out after a data migration is activated but is not complete
򐂰 Back out after a data migration reaches the synchronized state
6.10.1 Back out before migration is defined on the XIV

If a data migration definition does not exist yet, then no action must be taken on the XIV. You
can zone the host server back to the non-XIV storage system and map LUNs on the host
server back to the host server. Make sure that the correct LUN order is preserved.
6.10.2 Back out after a data migration is defined but not activated
If the data migration definition exists but is not activated, follow the steps described in 6.10.1,
“Back out before migration is defined on the XIV” on page 169. To remove the inactive
migration from the migration list, delete the XIV volume that was going to receive the migrated
data.
6.10.3 Back out after a data migration is activated but is not complete
If the data migration status is initialization in the GUI or the XCLI shows it as active=yes, the
background copy process is started. Deactivating the migration in this state blocks any I/O
passing through the XIV from the host server to the LUNs on the XIV and non-XIV systems.
To back out, first shut down the host server or its applications. Then deactivate the data
migration and delete the XIV data migration volume if wanted. Finally, restore the original LUN
masking and SAN fabric zoning and bring your host back up.
Important: Choosing to not allow source updating and write I/O occurs after the migration
started prevents the LUN on the non-XIV storage device from containing those writes.

6.10.4 Back out after a data migration reaches the synchronized state
If the data migration status in the GUI is synchronized, the background copy is complete. In
this case backout can still occur because the data migration is not destructive to the source
LUN on the non-XIV storage device. Reverse the process by shutting down the host server or
applications and restoring the original LUN masking and switch zoning settings. You might
need to also reinstall the relevant host server multi-path software for access to the non-XIV
storage device.
Important: If you chose to not allow source updating and write I/O has occurred during or
after the migration, the LUN on the non-XIV storage device will not contain those writes.
6.11 Migration checklist

There are three separate stages to a migration:
1. Prepare the environment for the implementation of the XIV
2. Cut over your hosts
3. Remove any old devices and definitions
For site setup, the high-level process includes the following steps:
1. Install XIV and cable it into the SAN.
2. Pre-populate SAN zones in switches.
3. Pre-populate the host/cluster definitions in the XIV.
4. Define XIV to the non-XIV disk as a host.
5. Define the non-XIV disk to XIV as a migration target.
6. Confirm paths.
For each host, the high-level process includes the following steps:
1. Update host drivers, install Host Attachment Kit, and shut down the host.
2. Disconnect/Un-Zone the host from non-XIV storage.
3. Zone the host to XIV.
4. Map the host LUNs away from the host instead of mapping them to the XIV.
5. Create XIV data migration.
6. Map XIV data migration volumes to the host.
7. Start the host.
When all data on the non-XIV disk system is migrated, perform site cleanup using these
steps:
1. Delete all SAN zones related to the non-XIV disk.
2. Delete all LUNs on non-XIV disk and remove it from the site.

Table 6-1 shows the site setup checklist.
Table 6-1 Physical site setup

Task Completed Where to Task
Number perform
1 Site Install XIV.
2 Site Run fiber cables from SAN switches to XIV for host connections
and migration connections.
3 Non-XIV storage Select host ports on the non-XIV storage to be used for migration
traffic. These ports do not have to be dedicated ports. Run new
cables if necessary.
4 Fabric switches Create switch aliases for each XIV Fibre Channel port and any
new non-XIV ports added to the fabric.
5 Fabric switches Define SAN zones to connect hosts to XIV, but do not activate the
zones. Define them by cloning the existing zones from host to
non-XIV disk and swapping non-XIV aliases for new XIV aliases.
6 Fabric switches Define and activate SAN zones to connect non-XIV storage to XIV
initiator ports (unless direct connected).
7 Non-XIV storage If necessary, create a small LUN to be used as LUN0 to allocate

to the XIV.
8 Non-XIV storage Define the XIV on the non-XIV storage device, mapping LUN0 to
test the link.
9 XIV Define non-XIV storage to the XIV as a migration target and add
ports. Confirm that links are green and working.
10 XIV Change the max_initialization_rate depending on the non-XIV

disk. You might want to start at a smaller value and increase it if
no issues are seen.
11 XIV Define all the host servers to the XIV (cluster first if using clustered
hosts). Use a host listing from the non-XIV disk to get the WWPNs
for each host.
12 XIV Create storage pools as required. Ensure that there is enough

pool space for all the non-XIV disk LUNs being migrated.
After the site setup is complete, the host migrations can begin. Table 6-2 shows the host
migration checklist. Repeat this checklist for every host. Task numbers identified with an
asterisk must be performed with the host application offline.
Table 6-2 Host Migration to XIV task list

Task Completed? Where to Task
number perform
1 Host From the host, determine the volumes to be migrated and their relevant LUN
IDs and hardware serial numbers or identifiers.
2 Host If the host is remote from your location, confirm that you can power the host
back on after shutting it down. You can use tools such as an RSA card or IBM
BladeCenter® manager.
3 Non-XIV Get the LUN IDs of the LUNs to be migrated from non-XIV storage device.
Storage Convert from hex to decimal if necessary.

number perform
4* Host Shut down the application.
5* Host Set the application to not start automatically at reboot. Using this setting helps
when performing administrative functions on the server such as upgrades of
drivers, patches, and so on.
6* Host On UNIX servers, comment out disk mount points on affected disks in the
mount configuration file. Commenting them out helps with system reboots while
configuring for XIV.
7* Host Shut down affected servers.
8* Fabric Change the active zone set to exclude the SAN zone that connects the host
server to non-XIV storage. In addition, include the SAN zone for the host server
to XIV storage. Create the new zone during site setup.
9* Non-XIV Unmap source volumes from the host server.

storage
10* Non-XIV Map source volumes to the XIV host definition created during site setup.
storage
11* XIV Create data migration pairing (XIV volumes created dynamically).
12* XIV Test XIV migration for each volume.
13* XIV Start XIV migration and verify it. If you want, wait for migration to finish.
14* Host Boot the server. Be sure that the server is not attached to any storage.
15* Host Coexistence of non-XIV and XIV multi-pathing software is supported with an
approved SCORE(RPQ) only. Remove any unapproved multi-pathing software
16* Host Install patches, update drivers, and HBA firmware as necessary.
17* Host Install the XIV Host Attachment Kit. Be sure to note the prerequisites.
18* Host You might need to reboot depending on the operating system.
19* XIV Map XIV volumes to the host server. (Use original LUN IDs.)
20* Host Reboot server
21* Host Verify that the LUNs are available and that pathing is correct.
22* Host For UNIX servers, update the mount points for new disks in the mount
configuration file if they have changed. Mount the file systems.
23* Host Start the application.
24* Host Set the application to start automatically if this setting was previously changed.
25 XIV Monitor the migration if it is not already completed.
26 XIV When the volume is synchronized, delete the data migration. Do not deactivate
the migration.
27 Non-XIV Unmap migration volumes away from XIV if you need to free up LUN IDs.
Storage
28 XIV Consider resizing the migrated volumes to the next 17 GB boundary if the host
operating system is able to use new space on a resized volume.

number perform
29 Host If XIV volume was resized, use host procedures to use the extra space.
30 Host If non-XIV storage device drivers and other supporting software were not
removed earlier, remove them when convenient.
When all the hosts and volumes are migrated, perform the site cleanup tasks shown in
Table 6-3.
Table 6-3 Site cleanup checklist

number perform
1 XIV Delete migration paths and targets.
2 Fabric Delete all zones related to non-XIV storage,

including the zone for XIV migration.
3 Non-XIV Delete all LUNs and perform secure data

storage destruction if required.
6.12 Other considerations

The XIV supports migration from practically any SCSI storage device that has Fibre Channel
interfaces. To ensure a successful migration from your specific storage device to XIV, the
following items must be validated:
򐂰 LUN0: Do you need to specifically map a LUN to LUN ID zero? This determines whether
you will have a problem defining the paths.
򐂰 LUN numbering: Does the storage device GUI or CLI use decimal or hexadecimal LUN
numbering? This determines whether you must do a conversion when entering LUN
numbers into the XIV GUI.
򐂰 Multipathing: Is the device active/active or active/passive? This determines whether you
define the storage device as a single target, or as one target per internal controller or
service processor.
򐂰 Definitions: Does the device have specific requirements when defining hosts?
For more information, see IBM XIV Storage System: Copy Services and Migration,
SG24-7759, at:
http://www.redbooks.ibm.com/redbooks/pdfs/sg247759.pdf

7
Chapter 7. SAN Volume Controller-based

migration
This chapter addresses various methods for migrating data from any SAN Volume Controller
supported storage subsystem to a different one. The storage types covered include between
single or multiple, and similar or dissimilar.
This chapter provides a brief description of the IBM System Storage SAN Volume Controller,
SAN Volume Controller terminology, and architecture.
In October 2010, IBM introduced a new midrange disk system called IBM Storwize® V7000.
V7000 is based on SAN Volume Controller virtualization technology and provides the same
functions and interoperability as SAN Volume Controller.
For more information about Storwize, see the following web address:
http://www-03.ibm.com/systems/storage/disk/storwize_v7000/

򐂰 IBM System Storage SAN Volume Controller overview
򐂰 SAN Volume Controller concepts for migrating the data
򐂰 Migrating using SAN Volume Controller
򐂰 SAN Volume Controller Migration preparation prerequisites
򐂰 Migrating SAN disks to SAN Volume Controller volumes and back to SAN
򐂰 SAN Volume Controller Volume migration between two storage pools
򐂰 Data migration using SAN Volume Controller mirrored volumes
򐂰 Data migration using SAN Volume Controller Metro Mirror
򐂰 SAN Volume Controller as data migration engine
򐂰 Other resources

7.1 IBM System Storage SAN Volume Controller overview
The IBM System Storage SAN Volume Controller is a scalable hardware and software
solution. It provides block aggregation and logical drive management for different disk storage
subsystems in a SAN environment. SAN Volume Controller provides the following
advantages:
򐂰 A single view of the storage attached to the SAN: You can manage, add, and migrate
physical disks non-disruptively, even between storage subsystems.
Tip: The SAN must be zoned in such a way that the application servers cannot see the
storage. This zoning prevents conflict between the SAN Volume Controller and the
application servers that are both trying to manage the storage.
򐂰 Storage virtualization: SAN Volume Controller creates a storage pool of managed disks
from attached disk storage subsystems. These managed disks are then mapped to a set
of volumes for use by host computer systems.
򐂰 Scalable: SAN Volume Controller can be used to manage all of your disk storage
requirements, or just a subset of them. SAN Volume Controller also offers a large scalable
cache using an algorithm.
򐂰 Reduces the requirement for additional partitions: SAN Volume Controller consumes only
one storage partition for each storage server that connects to it.
򐂰 Improves access: SAN Volume Controller improves capacity utilization, and spare
capacity. Underlying physical disks can be reallocated non-disruptively from an application
server point of view irrespective of the server operating system or platform type.
򐂰 Simplifies device driver configuration on hosts: All hosts within your network use the same
IBM device driver to access all storage subsystems through the SAN Volume Controller.
򐂰 Supports split I/O group implementations: Split implementations are not apparent to an
application and are used across sites to cover application high availability requirements.
These implementations can be used in case of high availability demands.
SAN Volume Controller is licensed according to the usable capacity that is being managed.
The advanced functions available on SAN Volume Controller, such as FlashCopy, IBM Easy
Tier®, Split I/O group, Mirrored volumes, Metro Mirror, and Global Mirror are included.
The license cost is for the capacity of all storage managed by the SAN Volume Controller,
plus the capacity of the copy services maintained by the SAN Volume Controller. You can
upgrade at any time by purchasing a license for the additional capacity required.
SAN Volume Controller supports a wide variety of disk storage and host operating system
platforms. For the latest information, see the following web address:
http://www-03.ibm.com/systems/storage/software/virtualization/svc/interop.html
For details and information about SAN Volume Controller implementation, see Implementing
the IBM System Storage SAN Volume Controller V6.1, SG24-7933.

7.1.1 SAN Volume Controller components, concepts, and terminology
An SAN Volume Controller implementation consists of both hardware and software. The
hardware consists of the following items:
򐂰 A management console
򐂰 A minimum of two node pairs to form the SAN Volume Controller cluster
򐂰 A minimum of two uninterruptible power supplies (UPSs)
According to new hardware and software releases (at the writing of this book, release 6.2),
SAN Volume Controller supports the following additional new functions:
򐂰 Support for 10 Gbps iSCSI functionality with 2145-CG8 node hardware
򐂰 Real-time performance monitoring
򐂰 Support for VMware vStorage APIs for Array Integration (VAAI)
򐂰 Internal SSD drive support
򐂰 Support for FlashCopy targets as Metro or Global Mirror sources
򐂰 Critical update notifications
򐂰 Additional language support
򐂰 New management GUI
򐂰 No requirement for separate console installation
򐂰 Direct cluster access using a web browser
򐂰 Easy Tier hotspot management
򐂰 Service Assistant
Remember: Depending on your SAN Volume Controller code level, some functions might
not be available.
SAN Volume Controller naming terminology has recently changed as shown in Table 7-1.
Table 7-1 Recent naming changes

SAN Volume Controller 5.1 SAN Volume Controller 6.1 and
later
Managed Disk Group Storage pool
Virtual Disk (VDisk) Volume
Space-efficient Thin provisioning
VDisk-to-host mapping Host mapping
Error Event
Chapter 7. SAN Volume Controller-based migration 177

SAN Volume Controller management
With the new SAN Volume Controller code release, you can access SAN Volume Controller
nodes directly through a web browser or using the command-line interface (CLI). It is no
longer necessary to have an SAN Volume Controller Console installed. The web GUI is
embedded in the SAN Volume Controller code and can be access directly from any
workstation using a web browser. The IBM SSPC (System Storage Productivity Center)
server is still in place and is valuable as central point of management.
Node
A node is an individual server in a SAN Volume Controller cluster on which the SAN Volume
Controller software runs. Nodes are always installed as pairs and represent one I/O group in
the SAN Volume Controller cluster concept. Each node is connected to its own UPS to ensure
a safe power off during a power outage.
Configuration node
At any time, a single node in the cluster is used to manage configuration activity. This
configuration node manages an information cache that describes the cluster configuration
and provides a focal point for configuration commands.
Similarly, at any time, a single node is responsible for the overall management of the cluster.
In a configuration node failure, the SAN Volume Controller fails over to the second node,
which takes the configuration function. The IP address and user credentials remain the same.
SAN Volume Controller cluster

When the first two nodes are installed, they form an SAN Volume Controller cluster that can
contain up to eight nodes (four pairs of nodes).
I/O group
An input/output (I/O) group contains two SAN Volume Controller nodes defined by the
configuration process. Each SAN Volume Controller node is associated with exactly one I/O
group. The nodes in the I/O group provide access to the volumes in the I/O group.
Internal storage and external storage

SAN Volume Controller system can manage a combination of internal and external storage:
򐂰 Internal storage: SAN Volume Controller model 2145-CF8 and model 2145-CG8 support
up to eight SSD drives attached to them per I/O group. These drives are used to create a
Redundant Array of Independent Disks (RAID) that is presented as managed disks
(MDisks) to the system.
򐂰 External storage: The SAN Volume Controller can detect logical units (LUs) on an external
storage system that is attached through Fibre Channel connections. These LUs are
detected as managed disks (MDisks) in the system, and must be protected from drive
failures using RAID technology on the storage system.
Front-End and back-end

SAN Volume Controller takes managed disks from back-end storage subsystem, groups them
in storage pools, virtualizes them, and presents them to application servers (hosts) as
volumes.
The switch fabric must be zoned, so that the SAN Volume Controller nodes can detect the
back-end storage systems and the front-end host HBAs. The SAN Volume Controller prevents
direct host access to managed disks. The managed disks are looked after by the back-end

application of the SAN Volume Controller. The volumes presented to hosts are looked after by
the front-end application in the SAN Volume Controller.
Managed disks (MDisks)

A managed disk (MDisk) is a disk presented by a storage subsystem and managed by the
SAN Volume Controller. This MDisk must be on a RAID storage subsystem. An MDisk
provides usable blocks (or extents) of physical storage to the SAN Volume Controller cluster.
The MDisks must be zoned in a way that they are not visible to host systems on the SAN. An
MDisk in the SAN Volume Controller can be either managed or unmanaged. A managed
MDisk is an MDisk assigned to a storage pool.
Storage pool
A storage pool is a collection of MDisks that jointly contain all the data for a specified set of
volumes. Each storage pool is divided into a number of extents. When creating a storage pool,
you must choose an extent size. After it is set, the extent size stays constant for the life of that
storage pool. Each storage pool can have a different extent size.
Volumes
A volume is a logical entity that represents extents contained in one or more MDisks from a
storage pool. Volumes are allocated in whole numbers of extents. Each volume is only
associated with a single I/O group. The volume is then presented to the host as a LUN for
use.
There are three types of volumes:

򐂰 Striped: A volume created in striped mode has extents allocated from each MDisk in the
storage pool in a round-robin fashion.
򐂰 Sequential: The volumes are allocated sequentially on one MDisk to create the volume.
򐂰 Image: Image-mode volumes are special volumes that have a direct relationship with one
MDisk. If you have a Mdisk that contains data that you want to merge into the clustered
system, you can create an image-mode volume.
Easy Tier function

SAN Volume Controller includes Storage Easy Tier, a function that responds to the presence
of solid-state drives (SSDs) in a storage pool that also contains hard disk drives (HDDs). The
system automatically and non-disruptively moves frequently accessed data from HDD
MDisks to SSD MDisks, placing the data in a faster tier of storage.
In this dynamically tiered environment, data movement is seamless to the host application
regardless of the storage tier in which the data is located.
Quorum disks
Quorum disks are used when there is a problem in the SAN fabric or when nodes are shut
down. SAN Volume Controller uses quorum disks in such situation to ensure data
consistency and data integrity while maintaining data access. A quorum disk determines
which group of nodes stops operating and processing I/O requests. In this tie-break situation,
the first group of nodes that accesses the quorum disk marks its ownership of the quorum
disk. As a result, the group continues to operate as the system, handling all I/O requests.
Access modes
The access mode determines how the SAN Volume Controller system uses the MDisk. The
following are the types of access modes:

򐂰 Unmanaged: The MDisk is not used by the system.
򐂰 Managed: The MDisk is assigned to a storage pool and provides extents that volumes can
use.
򐂰 Image: The MDisk is assigned directly to a volume with a one-to-one mapping of extents
between the MDisk and the volume.
򐂰 Array: The MDisk represents a set of drives in a RAID from internal storage.
Important: If you add an MDisk that contains existing data to a storage pool while the
MDisk is in unmanaged or managed mode, the data is lost. The image mode is the only
mode that preserves this data.
Mirrored volumes
Volume mirroring allows a volume to have two physical copies. The mirrored volume feature
provides a simple RAID-1 function. Each volume copy can belong to a different storage pool,
and can belong to separate storage subsystems.
The primary copy indicates the preferred volume for read requests. When a server writes to a
mirrored volume, the system writes the data to both copies. You can create a volume with one
or two copies, and you can convert a non-mirrored volume into a mirrored volume by adding a
copy. Alternately, you can convert a mirrored volume into a non-mirrored volume by deleting
one copy or by splitting one copy to create a new, non-mirrored volume.
System state
The state of the clustered system holds all of the configuration and internal data. The system
state information is held in nonvolatile memory. If the power fails, the UPS units maintain
internal power long enough for system state information to be stored. This information is
stored on the SAN Volume Controller internal disk drive of each node.
Back up and restore process

Planning your backup/restore procedures is an important part of a complete backup and
disaster recovery solution.
Backup is the process of extracting configuration settings from a SAN Volume Controller
system and writing them to disk. The restore process uses backup configuration data files to
restore system configuration.
If power fails on a system or a node in a system is replaced, the system configuration settings
are automatically restored. This restoration occurs when the repaired node is added to the
system. To restore the system configuration in a disaster, plan to back up the system
configuration settings to tertiary storage. Use the configuration backup functions to back up
the system configuration.
Important: For complete disaster recovery, regularly back up the business data that is
stored on volumes at the application server or host level.
Event notifications
SAN Volume Controller can use Simple Network Management Protocol (SNMP) traps, syslog
messages, and call home email. These alerts notify you and the IBM Support Center when
significant events are detected.
These notification methods can be used simultaneously. Notifications are normally sent
immediately after an event occurs.

SAN Volume Controller migration
The SAN Volume Controller allows you to migrate extents that belong to a volume from
source managed disk (MDisk) to a target MDisk. It can do so without interrupting host access
to the associated volume. This function is used when performing volume migrations, and can
be performed for any volume defined on the SAN Volume Controller.
SAN Volume Controller migration can be used for the following tasks:
򐂰 Redistribution of volumes and their workload within an SAN Volume Controller cluster
across back-end storage
򐂰 Moving workload onto newly installed storage subsystems
򐂰 Moving workload off storage so that old or failing storage subsystems can be
decommissioned
򐂰 Moving workload to rebalance a changed workload
򐂰 Migrating data from earlier model back-end storage to SVC-managed storage
򐂰 Migrating data from one back-end controller to another using the SAN Volume Controller
as a data block mover
򐂰 Removing the SAN Volume Controller from the SAN
򐂰 Migrating data from managed mode back into image mode before removing the SAN
Volume Controller from a SAN
SAN Volume Controller migration can be performed at either the volume or the extent level,
depending on the purpose of the migration. The following are the supported migration
activities:
򐂰 Migrating extents within a storage pool and redistributing the extents of a volume on the
MDisks in the storage pool
򐂰 Migrating extents off an MDisk to other MDisks in the storage pool so the MDisk can be
removed
򐂰 Migrating a volume from one storage pool to another
򐂰 Changing the virtualization type of the volume to Image mode
򐂰 Migrating a volume between I/O groups
Extents
An extent is a fixed-size unit of data that is used to manage the mapping of data between
MDisks and volumes. The extent size choices are 16, 32, 64, 128, 256, 512, 1024, 2048,
4096, and 8192 MB. The choice of extent size affects the total storage that can be managed
by the SAN Volume Controller. For the 8192-MB extent size, SAN Volume Controller supports
the following capacities:
򐂰 A maximum volume capacity of 256 TB
򐂰 A maximum SAN Volume Controller system capacity of 32 PB

Figure 7-1 shows the relationship between extents and a volume.
Figure 7-1 Extents used to create a volume
Figure 7-2 shows the relationship between physical and virtual disks.
Mapping to Hosts
w/SDD or supported MultiPath Driver
Space-efficient
Volume vdisk0 vdisk1 vdisk2 vdisk3 vdisk4 vdisk5

125GB 10GB 525GB 1500GB 275GB 5GB
Storage
Pool mdiskgrp0 [EMC Group] mdiskgrp1 [IBM Group]
Stripe 400GB 600GB
16MB –8GB
Managed
Disk mdisk0 mdisk1 mdisk2 mdisk3 mdisk4 mdisk5 mdisk6
100GB 100GB 100GB 100GB 200GB 200GB 200GB
LUN OEM OEM OEM OEM IBM IBM IBM

100GB 100GB 100GB 100GB 200GB 200GB 200GB
Figure 7-2 Relationships between physical and virtual disks
Consistency groups
Consistency groups preserve data consistency across multiple volumes, especially for
applications that have related data that spans multiple volumes. To preserve the integrity of
the data being written, ensure that dependent writes are run in the intended sequence of the
application.
Multipath Subsystem Device Driver (SDD)

For resilience, availability, redundancy, and throughput reasons, use multipath device drivers.
IBM Multipath Subsystem Device Driver (SDD) is available for most operating system
platforms. In addition, some vendors provide their own.

See the IBM SAN Volume Controller website for the latest device drivers and the support
matrix at:
http://www.ibm.com/storage/support/2145
SDD software for all supported platforms can be obtained for no additional fee at:
https://www-304.ibm.com/support/docview.wss?uid=ssg1S7001350
7.1.2 SAN Volume Controller copy services

The advanced function copy services offered by the SAN Volume Controller are FlashCopy,
Metro Mirror, and Global Mirror. If you plan to use SAN Volume Controller for all your copy
services needs, you might not need to purchase additional premium copy services features.
The SAN Volume Controller copy services functions can provide additional capabilities and
unique advantages over other storage devices:
򐂰 SAN Volume Controller supports consistency groups for FlashCopy, Metro Mirror, and
Global Mirror.
򐂰 Consistency groups in SAN Volume Controller can span across underlying storage
subsystems.
򐂰 FlashCopy source volumes that are on one disk system can write to target volumes on
another disk system.
򐂰 SAN Volume Controller supports FlashCopy targets as Metro or Global Mirror sources
(SAN Volume Controller code 6.2.0.1)
򐂰 Metro Mirror and Global Mirror source volumes can be copied to target volumes on a
dissimilar storage subsystem.
7.1.3 Metro Mirror

The Metro Mirror copy service is a synchronous remote copy function. Metro Mirror in SAN
Volume Controller is similar to Metro Mirror in the IBM System Storage DS® family.
The function of Metro Mirror is to maintain two real-time synchronized copies of a data set.
Often, the two copies are geographically dispersed on two SAN Volume Controller clusters,
although you can use Metro Mirror in a single cluster (within an I/O group). If the primary copy
fails, the secondary copy can then be enabled for I/O operation.
Metro Mirror works by defining a Metro Mirror relationship between volumes of equal size.
When creating the Metro Mirror relationship, define one volume as the master, and the other
volume as the auxiliary. Any data that exists on the auxiliary volume before the relationship
being set up is deleted.
The terms master and auxiliary are used instead of the industry standard terms of source
and target. These terms are used because they match the parameter keywords used in the
commands to set up the relationships between the volumes. The following terms are all
interchangeable:
򐂰 Source and target
򐂰 Primary and secondary
򐂰 Master and auxiliary
To provide management and data consistency across a number of Metro Mirror relationships,
consistency groups are supported.

The SAN Volume Controller provides both intracluster and intercluster Metro Mirror as
described in the following sections. The examples of migration in this chapter use two SAN
Volume Controller clusters, which is intercluster Metro Mirroring.
Intracluster Metro Mirror

Intracluster Metro Mirror can be applied within any single I/O group. Metro Mirror across I/O
groups in the same SAN Volume Controller cluster is not supported.
Intercluster Metro Mirror

A Metro Mirror session is a form of synchronous remote replication, designed to operate over
distances under 300 kilometers (km). It maintains identical data in both the source and target.
Changes made to the source data are propagated to the target before the write finishes
posting. With Metro Mirror, the source is located in one subsystem and the target is located in
another subsystem.
Intercluster Metro Mirror operations require a pair of SAN Volume Controller clusters that are
separated by a number of moderately high-bandwidth links. The two SAN Volume Controller
clusters must be defined in an SAN Volume Controller partnership. This definition must be
performed on both SAN Volume Controller clusters to establish a fully functional Metro Mirror
partnership. Correct link sizing is crucial to successfully implement Metro Mirror replication.
Metro Mirror is a fully synchronous remote copy technique. It ensures that updates (writes),
are committed at both primary and secondary volumes before the application is given “write
successful” status to an update and released. A write to the master volume is mirrored to the
cache for the auxiliary volume. An acknowledge of the write is then sent back to the host
(Figure 7-3).
Figure 7-3 Remote Mirroring synchronous write sequence
If your system goes down, Metro Mirror provides nearly zero loss if data must be used from
the recovery site. While the Metro Mirror relationship is active, the secondary copy (auxiliary
volume) is not accessible for host application write I/O at any time. The SAN Volume
Controller allows read-only access to the secondary volume, when it contains a consistent
image. To allow access to the secondary volume for host operations, the Metro Mirror
relationship must first be stopped.

7.1.4 Global Mirror
Global Mirror (GM) works by defining a GM relationship between two volumes of equal size
and maintains the data consistency in an asynchronous manner.
Because SAN Volume Controller Global Mirror is not intended to use for data migration, it is
not addressed further here. For more information about SAN Volume Controller Global Mirror,
see Software Installation and Configuration Guide, GC27-2286-01.
7.2 SAN Volume Controller concepts for migrating the data

This section explains how to use SAN Volume Controller as a migration tool for block-level
(LUN) migrations between storage subsystems. SAN Volume Controller is a robust and
reliable data migration solution. When addressing complexity and data integrity, SAN Volume
Controller in usually the best solution for data migration. The following SAN Volume Controller
methods are available to use in migration scenarios:
򐂰 Data migration using SAN Volume Controller volume migration
򐂰 Data migration using SAN Volume Controller FlashCopy
򐂰 Data migration using SAN Volume Controller Metro Mirror
򐂰 Data migration using mirrored volumes
7.2.1 Data migration using SAN Volume Controller volume migration

SAN Volume Controller Volume migration combines migrating a volume to an image mode
volume, with the ability to migrate between storage pools. The source for the migration is an
image mode VDisk.
To be able to migrate, the destination MDisk must be greater than or equal to the size of the
VDisk. Also, the MDisk specified as the target must be in an unmanaged state at the time the
command is run.
If the migration is interrupted by a cluster recovery, the migration will resume after the
recovery completes.
Volume migration algorithm

The following section describes the algorithm for the volume migration.
Regardless of the extent size for the Storage pool, data is migrated in units of 16 MB. In this
description, this unit is called a chunk.
The algorithm to migrate an extent includes the following steps:

1. Pause all I/O on the source MDisk on all nodes in the SAN Volume Controller cluster.
During the pause, the specific mdisk SVC stores all I/O in the virtualization layer and waits
for all outstanding requests to complete. Therefore, the host is not aware that the migration
is underway. The I/O to other extents is unaffected.
2. Unpause I/O on all of the source MDisk extents other than writes to the specific chunk that
is being migrated. Writes to the extent are mirrored to the source and destination.
3. On the node performing the migration, for each 256K section of the chunk:
– Synchronously read 256K from the source.
– Synchronously write 256K to the target.

4. After the entire chunk is copied to the destination, repeat the process for the next chunk in
the extent.
5. After the entire extent is migrated, perform the following steps:
a. Pause all I/O to the extent being migrated
b. Checkpoint the extent move to on-disk metadata
c. Redirect all further reads to the destination
d. Stop mirroring writes (writes only to the destination)
If the checkpoint fails, the I/O is unpaused.
During the migration, the extent can be divided into three regions as shown in Figure 7-4.
Region B is the chunk that is being copied. The reads and writes are in the following states:
򐂰 Writes to region B are queued (paused) in the virtualization layer, waiting for the chunk to
be copied.
򐂰 Reads to Region A are directed to the destination because this data has already been
copied.
򐂰 Writes to Region A are written to both the source and the destination extent to maintain
the integrity of the source extent.
򐂰 Reads and writes to Region C are directed to the source because this region has yet to be
migrated.
Managed Disk Extents
Extent N-1 Extent N Extent N+1
Region A Region B Region C

(already copied) (copying) (yet to be copied)
reads/writes go reads/writes reads/writes go to
to destination paused source
Not to scale
16 MB
Figure 7-4 Migrating an extent
The migration of a chunk requires 64 synchronous reads and 64 synchronous writes. During
this time, all writes to the chunk from higher layers in the software stack (such as cache
destages) are held back. If the back-end storage is operating with significant latency, this
operation might take some time to complete. This latency can have an adverse effect on the
overall performance of the SAN Volume Controller. To avoid this situation, if the migration of a
particular chunk is still active after one minute, the migration is paused for 30 seconds. During
this time, writes to the chunk are allowed to proceed. After 30 seconds, the migration of the
chunk is resumed. This process is repeated as many times as necessary to complete the
migration of the chunk.

Considerations for host I/O during volume migration
SAN Volume Controller guarantees read stability during data migrations, even if the data
migration is stopped by a node reset or a cluster shutdown. SAN Volume Controller disallows
writes on all nodes to the area being copied, and upon a failure the extent migration is
restarted from the beginning.
However, SAN Volume Controller does not guarantee data synchronization on writes during
the migration. After the metadata is updated for each 16-MB chuck, any new writes are
redirected to the target disk only. Because of the nature of the migration algorithm, restarting
hosts in the event of an SAN Volume Controller cluster failure is difficult.
Important: To guarantee that the source and target data is synchronous using this method
of migration, pause I/O during the migration. The risk of a complete SAN Volume Controller
cluster failure during migration is small. Weigh the decision to pause I/O during migration
against the impact of any I/O downtime.
7.2.2 Data migration using SAN Volume Controller FlashCopy

The FlashCopy function allows you to create a rapid, complete, and consistent copy from a
source volume to a target volume. This feature can be used effectively for migrating data to
and from SAN Volume Controller managed storage.
When SAN Volume Controller is introduced, it takes time and effort to virtualize the storage.
Migration of data from non-virtualized to virtualized storage can be achieved in many different
ways. Image to managed volume migration and volume mirroring are the traditional
approaches for data migration.
FlashCopy can also be used for migrating data from non-virtualized to SAN Volume Controller
managed storage. The target volume can be anywhere within the SVC-managed
environment. The typical scenario for the migration is from image volume to managed
volume. FlashCopy allows you to create a read/write copy on the target volume in a matter of
a few seconds.
The significant advantage of FlashCopy migration is that the source volume remains
unchanged throughout the migration process. Another advantage is that when the copy is
created, the target volume can instantly be used in read/write mode.
For details and information about SAN Volume Controller Copy Services, see SAN Volume
Controller V4.3.0 Advanced Copy Services, SG24-7574.
FlashCopy -based data migration provides a simple and reliable backout path. If something
did not go as planned during the migration, unmask the original volumes from the SAN
Volume Controller and mask them back to a server. You do not need to copy data back
because the original volumes remained unchanged.
You can also repeat the copy process from the source to the target as many times as
required. If for whatever reason the target set of volumes become unusable, you can refresh
the contents by recopying data.
The disadvantage of this method is that it requires slightly more effort compared to the
volume migration and volume mirroring methods. The reason for this is you need to create the
target managed volumes of matching sizes, and create and start one FlashCopy mapping per
volume pair. However, this process can be automated using migration scripts.

A FlashCopy based data migration process from non-virtualized to SAN Volume Controller
managed storage involves the following steps:
1. Stop applications on the server.
2. Uninstall the multipathing device driver.
3. Shut down the server.
4. Remove volume masking to the server.
5. Present volumes to SAN Volume Controller.
6. Discover MDisks on SAN Volume Controller.
7. Create Image mode volumes using the discovered MDisks.
8. Create matching size managed volumes.
9. Create FlashCopy mappings between source and target volumes.
10.Start the FlashCopy copying process.
11.Create and activate a zone between the server and SAN Volume Controller.
12.Create host mapping to present target volumes to the server.
13.Start the server.
14.Install the multipathing device driver.
15.Perform data validation for OS and applications.
16.Perform cleanup.
The FlashCopy mappings in step 9 can be created with -autodelete flag so that the mapping
is removed after background copy process is finished. By removing the mapping
automatically, you reduce the number of steps in the cleanup process in step 15.
After the FlashCopy mapping is created and the background copy process is started, monitor
the progress and adjust the copy rate to minimize the performance impact. In most cases, the
background copy has no measurable performance impact on the application workload. In this
case, you can adjust the rate so that the background copy can complete sooner. You can
schedule copy rate reduction, for example to minimize the performance impact during a
period of heavy batch processing.
In order for a data migration to be successful the set of volumes must be consistent. To
ensure volume consistency, all the volume manipulations must be done for the entire volume
set. You must not present a subset of volumes to be accessed by a server.
7.2.3 Data migration using SAN Volume Controller Metro Mirror

Data migrations using Metro Mirror are more complicated. Use two SAN Volume Controller
clusters (four nodes: each cluster must have a minimum of two nodes) when migrating using
Metro Mirror. You can use a single cluster Metro Mirror (intracluster Metro Mirror) for data
migration, but this configuration introduces a single point of failure. Any problems occurring in
this single SAN Volume Controller cluster or the fabric connecting the environment might
affect both the source and target data. This risk is greater before the source and target are
fully synchronized.
By using dual cluster Metro Mirroring (intercluster Metro Mirror), the source volumes receive
the host updates at all times during the data migration. If a problem occurs between the host
and its data volumes, a normal recovery is possible. In the background, the SAN Volume
Controller copies the data over to the target volumes. You can restart from the source or, if
necessary, from the target volumes.
The target LUN must be exactly the same size as the source LUN.

7.2.4 Data migration using mirrored volumes
Using volume mirroring creates two physical copies of a volume. Each volume copy can
belong to a separate storage pool, and each copy has the same virtual capacity as the
volume. In the management GUI, an asterisk (*) indicates the primary copy of the mirrored
volume. The primary copy indicates the preferred volume for read requests.
When a server writes to a mirrored volume, the system writes the data to both copies. When
a server reads a mirrored volume, the system picks one of the copies to read. If one of the
mirrored volume copies is temporarily unavailable, the volume remains accessible to servers
through the other copy. The system remembers which areas of the volume are written and
resynchronizes these areas when both copies are available.
You can create a volume with one or two copies, and you can convert a non-mirrored volume
into a mirrored volume by adding a copy. When a copy is added in this way, the SAN Volume
Controller clustered system synchronizes the new copy so that it is the same as the existing
volume. Servers can access the volume during this synchronization process.
You can convert a mirrored volume into a non-mirrored volume by deleting one copy or by
splitting one copy to create a new non-mirrored volume. The volume copy can be any type:
Image, striped, sequential, and either thin provisioned or fully allocated. The two copies can
be of different types.
You can use mirrored volumes for the following reasons:

򐂰 Improving availability of volumes by protecting them from a single storage system failure.
򐂰 Providing concurrent maintenance of a storage system that does not natively support
concurrent maintenance.
򐂰 Providing an alternative method of data migration with better availability characteristics.
While a volume is being migrated using the data migration feature, it is vulnerable to
failures on both the source and target storage pool. Volume mirroring starts with a
non-mirrored volume in the source storage pool, and then adds a copy to that volume in
the destination storage pool. When the volume is synchronized, you can delete the original
copy. During the synchronization process, the volume remains available even if there is a
problem with the destination storage pool.
򐂰 Converting between fully allocated volumes and thin-provisioned volumes.
When you use volume mirroring, consider how quorum candidate disks are allocated. Volume
mirroring maintains some state data on the quorum disks. If a quorum disk is not accessible
and volume mirroring is unable to update the state information, a mirrored volume might need
to be taken offline to maintain data integrity. To ensure the high availability of the system,
ensure that multiple quorum candidate disks, allocated on different storage systems, are
configured.

Figure 7-5 shows a logical layout of mirrored volumes.
Figure 7-5 SAN Volume Controller mirrored volumes
7.3 Migrating using SAN Volume Controller

This section explains how to migrate from a conventional storage infrastructure to a
virtualized storage infrastructure using the IBM System Storage SAN Volume Controller (SAN
Volume Controller). The section includes migrating between different types of MDisks. It also
addresses how to phase SAN Volume Controller out of a virtualized storage infrastructure.
You might want to phase it out, for example, after a trial period or after using the SAN Volume
Controller as a data migration tool. It includes examples of using Metro Mirror to migrate data,
and how to use SAN Volume Controller mirrored volume functions to migrate data from one
disk subsystem to another.
You can perform migration at either the volume or the extent level, depending on the purpose
of the migration. The following migration activities are supported:
򐂰 Migrating extents within a storage pool, redistributing the extents of a volume on the
MDisks within the same storage pool
򐂰 Migrating extents off an MDisk, which is removed from the storage pool, to other MDisks in
the same storage pool
򐂰 Migrating a volume from one storage pool to another
򐂰 Migrating a volume to change the virtualization type of the volume to image
򐂰 Migrating a volume between I/O groups

7.3.1 Migrating extents
The SAN Volume Controller provides various data migration features. These features can be
used to move data both within storage pools and between storage pools. These features can
be used concurrently with I/O operations. You can use either of these methods to migrate
data:
1. Migrating data (extents) from one MDisk to another within the same storage pool. This
method can be used to move data off highly used MDisks.
2. Migrating volumes from one storage pool to another. This method can be used to move
data off highly used storage pools.
You can determine which MDisks are heavily used by gathering and analyzing input/output
(I/O) statistics about nodes, MDisks, and volumes. Using this information, you can migrate
extents to less used MDisks in the same storage pool. This migration can only be performed
using the command-line tools.
If performance monitoring tools indicate that a managed disk in the pool is being overused,
migrate some of the data to other MDisks within the same storage pool.
1. Determine the number of extents that are in use by each volume for the MDisk using the
following CLI command:
lsmdiskextent mdiskname
2. From the number of extents that each volume is using on the MDisk, select some of them
to migrate elsewhere in the group.
3. Determine the storage pool that the MDisk belongs to using this CLI command:
lsmdisk mdiskname | ID
4. List the MDisks in the group by issuing the following CLI command:
lsmdisk -filtervalue mdisk_grp_name=mdiskgrpname
5. Select one of these MDisks as the target MDisk for the extents. You can determine how
many free extents exist on an MDisk using the CLI command:
lsfreeextents mdiskname
6. Issue the lsmdiskextent newmdiskname command for each of the target MDisks to ensure
that you are not just moving the over-utilization to another MDisk. Check that the volume
that owns the set of extents to be moved does not already own a large set of extents on
the target MDisk.
7. For each set of extents, issue this CLI command to move them to another MDisk:
migrateexts -source mdiskname | ID -exts [num_extents] -target newmdiskname |
ID -threads 4 vdiskid
Where [num_extents] is the number of extents on the vdiskid. The newmdiskname | ID
value is the name or ID of the MDisk to migrate this set of extents to.
Tip: The number of threads indicates the priority of the migration processing, where 1 is
the lowest priority and 4 is the highest priority.
8. Repeat the previous steps for each set of extents that you are moving.
You can check the progress of the migration by issuing this CLI command:
lsmigrate

7.3.2 Migrating extents off an MDisk that is being deleted
Before deleting an MDisk from a storage pool using the rmmdisk -force command, migrate
the extents on it onto other MDisks in the storage pool.
If a volume uses extents that need to be moved as a result of a rmmdisk command, the
virtualization type for that volume must be set to striped. This process is needed only if it was
previously sequential or image.
If the MDisk is operating in image mode, the MDisk transitions to managed mode while the
extents are being migrated. Upon deletion, it changes to unmanaged mode.
Remember: If the -force flag is not used and volumes occupy extents on one or more of
the MDisks that are specified, the command fails.
When the -force flag is used and volumes occupy extents on the specified MDisks, all
extents are migrated to other MDisks in the storage pool. This process will occur only if
there are enough free extents in the storage pool. The deletion of the MDisks is postponed
until all extents are migrated, which can take some time. If there are insufficient free
extents in the storage pool, the command fails.
7.3.3 Migrating a volume between storage pools

An entire volume can be migrated from one storage pool to another storage pool using the
migratevdisk command. A volume can be migrated between storage pools regardless of the
virtualization type (image, striped, or sequential), although it changes to striped. The
command you need to use varies depending on the type of migration, as shown in Table 7-2.
Table 7-2 Migration types and associated commands

Storage pool-to-storage pool type Command
Managed to managed migratevdisk
Image to managed migratevdisk
Managed to image migratetoimage
Image to image migratetoimage
Rule: The source and destination storage pool must have the same extent size for
migration to take place. Volume mirroring can also be used to migrate a volume between
storage pools if the extent sizes of the two pools are not the same.
Migration commands fail if the target or source volume is offline, or if there is insufficient
quorum disk space to store the metadata. Correct the offline or quorum disk condition and
reissue the command.
You can migrate volumes between storage pools using the command-line interface (CLI).
Determine the usage of particular MDisks by gathering input/output (I/O) statistics about
nodes, MDisks, and volumes. Analyze it to determine which volumes or MDisks need to be
moved to another storage pool.

Perform the following step to gather statistics about MDisks and volumes:
1. Use secure copy (scp) command to retrieve the dump files for analyzing. For example,
issue the following command, which copies all the volume statistics files to the AIX host in
the current directory:
scp clusterip:/dumps/iostats/v_*
2. Analyze the memory dumps to determine which volumes are overused. Determine which
MDisks are being used heavily so you can spread the data more evenly across all MDisks
in the group.
After you analyze the I/O statistics data, you can determine which volumes are hot. You also
need to determine the storage pool that you want to move this volume to. Either create a
storage pool or determine an existing group that is not yet heavily used. Check the I/O
statistics files that you generated for heavily used groups. Ensure that the MDisks or VDisks
in the target storage pool are used less than those in the source group.
You can use data migration or volume mirroring to migrate data between MDisk groups. Data
migration uses the command migratevdisk. Volume mirroring uses the commands
addvdiskcopy and rmvdiskcopy.
When you issue the migratevdisk command, a check is made to ensure that the destination
of the migration has enough free extents to satisfy the command. If it does not, the command
fails. This command takes some time to complete.
Perform the following steps to use the migratevdisk command to migrate volumes between
storage pools:
1. After you determine the volume that you want to migrate and the new storage pool you
want to migrate it to, issue the following CLI command:
migratevdisk -vdisk vdiskname/ID -mdiskgrp newmdiskgrname/ID -threads 4
2. You can check the progress of the migration by issuing the following CLI command:
lsmigrate
Note: When you use data migration, the volume goes offline if either storage pool fails.
Volume mirroring can be used to minimize the impact to the volume because the volume
goes offline only if the source storage pool fails
7.3.4 Using volume mirroring

Perform the following steps to use volume mirroring to migrate volumes between storage
pools:
1. After you determine the volume that you want to migrate and the new storage pool that you
want to migrate it to, issue the following command:
addvdiskcopy -mdiskgrp newmdiskgrname/ID vdiskname/ID
2. The copy ID of the new copy is returned. The copies now synchronize so the data is stored
in both storage pools.
3. You can check the progress of the synchronization by issuing the following command:
lsvdisksyncprogress

4. After the synchronization is complete, remove the copy from the original I/O group to free
up extents and decrease the utilization of the storage pool. To remove the original copy,
issue the following command:
rmvdiskcopy -copy original copy id vdiskname/ID
7.3.5 Image mode volume migration

This section addresses migrating data from an image mode volume to a fully managed
volume. This type of migration is used to take an existing host LUN and move it into the
virtualization environment as provided by the SAN Volume Controller.
MDisk modes
There are three MDisk modes:
򐂰 Unmanaged MDisk: An MDisk is reported as unmanaged when it is not a member of any
storage pool. An unmanaged MDisk is not associated with any volumes and has no
metadata stored on it. The SAN Volume Controller does not write to an MDisk that is in
unmanaged mode. The only exception is when it attempts to change the mode of the
MDisk to one of the other modes.
򐂰 Image mode MDisk: Image mode provides a direct block-for-block translation from the
MDisk to the volume with no virtualization. Image mode volumes have a minimum size of
one block (512 bytes) and always occupy at least one extent. An image mode MDisk is
associated with exactly one volume.
򐂰 Managed mode MDisk: Managed mode MDisks contribute extents to the pool of available
extents in the storage pool. Zero or more managed mode volumes might use these
extents.
Changing between the modes

The following state transitions can occur to an MDisk (Figure 7-6 on page 195):
򐂰 Unmanaged mode to managed mode: This transition occurs when an MDisk is added to a
storage pool, which makes the MDisk eligible for the allocation of data and metadata
extents.
򐂰 Managed mode to unmanaged mode: This transition occurs when an MDisk is removed
from a storage pool.
򐂰 Unmanaged mode to image mode: This transition occurs when an image mode MDisk is
created on an MDisk that was previously unmanaged. It also occurs when an MDisk is
used as the target for a migration to image mode.
򐂰 Image mode to unmanaged mode: There are two ways in which this transition can
happen:
– When an image mode volume is deleted, the MDisk that supported the volume
becomes unmanaged.
– When an image mode volume is migrated in image mode to another MDisk, the MDisk
that is being migrated from remains in image mode. When all data is moved off the
MDisk, it changes to unmanaged mode.

򐂰 Image mode to managed mode: This transition occurs when the image mode volume that
is using the MDisk is migrated into managed mode.
򐂰 Managed mode to image mode: There is no operation that takes an MDisk directly from
managed mode to image mode. You can achieve this transition by performing operations
that convert the MDisk to unmanaged mode and then to image mode.
Figure 7-6 Various states of a volume
Image mode volumes have the special property that the last extent in the volume can be a
partial extent. Managed mode disks do not have this property.
To perform any type of migration activity on an image mode volume, the image mode disk
must first be converted into a managed mode disk. If the image mode disk has a partial last
extent, this last extent in the image mode volume must be the first extent to be migrated. This
migration is handled as a special case. After the special migration operation occurs, the
volume becomes a managed mode volume. If the image mode disk does not have a partial
last extent, no special processing is needed. The image mode volume is changed into a
managed mode volume and is treated in the same way as any other managed mode volume.
After data is migrated off a partial extent, there is no way to migrate data back onto the partial
extent.
You can use the command-line interface (CLI) to import storage that contains existing data
and continue to use this storage. You can also use the advanced functions, such as Copy
Services, data migration, and the cache. These disks are known as image mode virtual
volumes.
Make sure that you are aware of the following restrictions before you create image mode
volumes:
1. Unmanaged-mode managed disks (MDisks) that contain existing data cannot be
differentiated from unmanaged-mode MDisks that are blank. Therefore, it is vital that you
control the introduction of these MDisks to the clustered system by adding these disks one
at a time.
2. Do not manually add an unmanaged-mode MDisk that contains existing data to a storage
pool. If you do, the data will be lost. Use the command to convert an image mode volume
from an unmanaged-mode disk, and select the storage pool you want it added to.

Perform the following steps to create an image mode volume:
1. Stop all I/O operations from the hosts.
2. Unmap the logical disks that contain the data from the hosts.
3. Create one or more storage pools.
4. Map a single array or logical unit from your RAID storage system to the clustered system.
You can do this mapping through a switch zoning or a RAID storage system based on your
host mappings. The array or logical unit is displayed as an unmanaged-mode MDisk to the
SAN Volume Controller.
5. Issue the lsmdisk command to list the unmanaged-mode MDisks. If the new
unmanaged-mode MDisk is not listed, perform a fabric-level discovery. Issue the
detectmdisk command to scan the Fibre Channel network for the unmanaged-mode
MDisks.
Tip: The detectmdisk command also rebalances MDisk access across the available
storage system device ports.
6. Convert the unmanaged-mode MDisk to an image mode virtual disk. Issue the mkvdisk
command to create an image mode virtual disk object.
7. Map the new volume to the hosts that were previously using the data that the MDisk now
contains.
You can use the mkvdiskhostmap command to create a mapping between a volume and a
host. This mapping makes the image mode volume accessible for I/O operations to the host.
After the volume is mapped to a host object, the volume is detected as a disk drive with which
the host can perform I/O operations.
If you want to virtualize the storage on an image mode volume, you can transform it into a
striped volume. Migrate the data on the image mode volume to managed-mode disks in
another storage pool. Issue the migratevdisk command to migrate an entire image mode
volume from one storage pool to another storage pool.
7.3.6 Migrating the volume to image mode

You can migrate a volume to an image mode volume and migrate between storage pools at
the same time. The source for the migration can be a managed mode or an image mode
volume. This leads to four possibilities:
򐂰 Migrate image mode-to-image mode within a storage pool.
򐂰 Migrate managed mode-to-image mode within a storage pool.
򐂰 Migrate image mode-to-image mode between storage pools.
򐂰 Migrate managed mode-to-image mode between storage pools.
The following conditions must apply before you can migrate:

򐂰 The destination MDisk must be greater than or equal to the size of the volume.
򐂰 The MDisk that is specified as the target must be in an unmanaged state at the time that
the command is run.
If the migration is interrupted by a cluster recovery, the migration will resume after the
recovery completes.

If the migration involves moving between storage pools, the volume behaves as described in
7.3.3, “Migrating a volume between storage pools” on page 192.
Regardless of the mode in which the volume starts, it is reported as being in managed mode
during the migration. In addition, both of the MDisks involved are reported as being in image
mode during the migration. Upon completion of the command, the volume is classified as an
image mode volume. Issuing this command results in the inclusion of the MDisk into the user
specified MDisk group.
The migratetoimage CLI command allows you to migrate the data from an existing VDisk
(volume) onto a managed disk (MDisk). When it is issued, it migrates the data of the user
specified source VDisk onto the specified target MDisk. When the command completes, the
VDisk is classified as an image mode VDisk.
Note: Migration commands fail if the target or source VDisk is offline, or if there is
insufficient quorum disk space to store the metadata. Correct the offline or quorum disk
condition and reissue the command.
Issue the following CLI command to migrate data to an image mode VDisk:
migratetoimage -vdisk [vdiskname/ID] -mdisk [newmdiskname/ID] -mdiskgrp
[newmdiskgrpname/ID] -threads 4
Where:
򐂰 [vdiskname/ID] is the name or ID of the VDisk
򐂰 [newmdiskname/ID] is the name or ID of the new MDisk
򐂰 [newmdiskgrpname/ID] is the name or ID of the new MDisk group (storage pool)
7.3.7 Migrating a volume between I/O groups

A volume can be migrated between I/O groups by using the svctask chvdisk command.
This command is only supported if the volume is not in a FlashCopy Mapping or Remote
Copy relationship.
To move a volume between I/O groups, the cache must first be flushed. The SAN Volume
Controller attempts to destage all write data for the volume from the cache during the I/O
group move. This flush fails if data is pinned in the cache for any reason, such as a storage
pool being offline. By default, this failed flush causes the migration between I/O groups to fail,
but this behavior can be overridden using the -force flag. If the -force flag is used and SAN
Volume Controller cannot destage all write data from the cache, the cached data is lost.
During the flush, the volume operates in cache write-through mode.
Important: Do not move a volume to an offline I/O group under any circumstance. Ensure
that the I/O group is online before you move the volumes to avoid any data loss.
You must quiesce host I/O before the migration for two reasons:
򐂰 If there is significant data in cache that takes a long time to destage, the command-line
times out.
򐂰 Subsystem Device Driver vpaths that are associated with the volume are deleted before
the volume move takes place to avoid data corruption. Therefore, data corruption can
occur if I/O is still occurring for a particular LUN ID.
When migrating a volume between I/O groups, you can specify the preferred node, or you can
allow SAN Volume Controller assign the preferred node.

A volume that is a member of a FlashCopy mapping or a Remote Copy relationship cannot be
moved to another I/O group. You cannot override this restriction using the -force flag. You
must delete the mapping or relationship before the volume can be migrated between I/O
groups.
Modifying the I/O group that services the volume cannot be done concurrently with I/O
operations. It also requires a rescan at the host level to ensure that the multipathing driver is
notified of the following conditions:
򐂰 The allocation of the preferred node has changed
򐂰 The ports by which the volume is accessed have changed.
Modify the group only when one pair of nodes becomes overused.
Perform the following steps to migrate a volume between I/O groups:

1. Synchronize all file systems that are mounted on the volume.
2. Stop all I/O operations to the volume.
3. Issue the following CLI command to migrate the volume into a new I/O group:
chvdisk -iogrp [iogrp_name_or_id] -node [preferred_node] [vdisk]
Where:
– [iogrp_name_or_id] is the name or ID of the I/O group that you want to migrate the
volume to
– [preferred_node] is the name of the node that you want to move the volume to
– [vdisk] is the name of the volume that you want to migrate
4. Resynchronize the volume to host mapping. For more information, see the IBM System
Storage Multipath Subsystem Device Driver User's Guide or the documentation provided
with your multipathing driver.
5. Restart the I/O operations to the volume.
7.3.8 Monitoring the migration progress

To monitor the progress of ongoing migrations, use the svcinfo lsmigrate CLI command. To
determine the extent allocation of MDisks and volumes, use the following commands.
򐂰 To list the volume IDs and the corresponding number of extents that the volumes occupy
on the queried MDisk, use the following CLI command:
svcinfo lsmdiskextent <mdiskname | mdisk_id>
򐂰 To list the MDisk IDs and the corresponding number of extents that the queried volumes
occupy on the listed MDisks, use the following CLI command:
svcinfo lsvdiskextent <vdiskname | vdisk_id>
򐂰 To list the number of available free extents on an MDisk, use the following CLI command:
svcinfo lsfreeextents <mdiskname | mdisk_id>
Important: After a migration is started, you cannot stop it. The migration runs to
completion unless it is stopped or suspended by an error condition, or the volume being
migrated is deleted.
If you want to start, suspend, or cancel a migration, or control the rate of migration, use the
volume mirroring function or migrate volumes between storage pools.

7.3.9 Parallelism
You can perform the following activities in parallel. The activities are divided into the following
categories:
򐂰 Per cluster
򐂰 Per MDisk
Per cluster
An SAN Volume Controller cluster supports up to 32 active concurrent instances of the
following migration activities:
򐂰 Migrate multiple extents
򐂰 Migrate between storage pools
򐂰 Migrate off a deleted MDisk
򐂰 Migrate to image mode
These high-level migration tasks can be started by scheduling single extent migrations. Up to
256 single extent migrations can run concurrently. This number can be any combination of the
migration activities previously listed.
The migrate multiple extents and migrate between storage pools commands support a
flag that allows you to specify the number of parallel “threads” to use. The flag can be set from
1 - 4. This parameter affects the number of extents that are concurrently migrated for that
migration operation. Thus, if the thread value is set to 4, up to four extents can be migrated
concurrently for that operation, subject to other resource constraints.
Per MDisk
The SAN Volume Controller supports up to four concurrent single extent migrates per MDisk.
This limit does not take into account whether the MDisk is the source or the destination. If
more than four single extent migrates are scheduled for a particular MDisk, further migrations
are queued pending the completion of the currently running migrations.
7.3.10 Error handling

The migration is suspended if any of the following conditions exist. If any other error is
encountered, it is stopped:
򐂰 The migration is between storage pools and has progressed beyond the first extent: These
migrations are always suspended because stopping a migration in progress leaves a
volume spanning storage pools. Spanning storage pools is not a valid configuration other
than during a migration.
򐂰 The migration is a Migrate to Image Mode, even if it is processing the first extent: These
migrations are always suspended rather than stopped because stopping a migration in
progress leaves the volume in an inconsistent state.
򐂰 A migration is waiting for a metadata checkpoint that has failed.
If a migration is stopped and migrations are queued awaiting the use of the MDisk, these
migrations now commence. However, if a migration is suspended, the migration continues to
use resources, and another migration is not started.
The SAN Volume Controller attempts to resume the migration if the error log entry is marked
as fixed using the CLI or the GUI. If the error condition no longer exists, the migration
proceeds. The migration might resume on a node other than the node that started the
migration.

7.3.11 Migration tips
Several methods are available to migrate an image mode volume to a managed mode
volume:
򐂰 If your image mode volume is in the same storage pool as the MDisks you want to migrate
to, perform one of these migrations:
– Migrate a single extent. You must migrate the last extent of the image mode volume
(number N-1).
– Migrate multiple extents.
– Migrate all of the in-use extents from an MDisk. For example, you must migrate extents
off an MDisk that is being deleted.
򐂰 If you have two storage pools (one for the image mode volume, and one for the managed
mode volumes), you can migrate a volume from one to the other.
Have one storage pool for all the image mode volumes, other storage pools for the managed
mode volumes, and use the migrate volume facility.
Be sure to verify that enough extents are available in the target storage pool.
7.4 SAN Volume Controller Migration preparation prerequisites

This section highlights important prerequisites steps to successfully implement the SAN
Volume Controller into an existing environment. This process includes the following steps:
򐂰 Fabric zoning
򐂰 Connect SAN Volume Controller to the fabric for migration
򐂰 Remove SAN Volume Controller from the fabric after migration
򐂰 Back-End storage consideration
򐂰 For the latest support information, see the SAN Volume Controller website at:
򐂰 Host attachment
7.4.1 Fabric zoning

This section addresses the basic steps for the following tasks:
򐂰 Inserting the SAN Volume Controller and the new storage into a fabric environment
򐂰 Making the changes needed when moving the SAN Volume Controller between the
storage and the servers
򐂰 Removing the SAN Volume Controller from the fabric
These steps assume that the SAN Volume Controller and storage are new to the fabric
environment.
Lay out new FC cables to the fabric switch to create the physical connections between the
target SAN Volume Controller cluster and the target storage. Have a clear picture of how you
want to zone the SAN Volume Controller to the back-end platform. Also, spread the fiber
adapter port connections for the most throughput and redundancy. Figure 7-7 on page 201
shows an example with two SAN Volume Controller nodes (one I/O group) and a back-end
disk system with eight ports.
The implementation can vary depending on back-end disk systems configuration. Always
check vendor recommendations for connectivity.

Figure 7-7 SAN Volume Controller connection diagram
Create new zones on the switch, to include the new systems. This zoning must be carefully
planned, to avoid any interruptions for other users of the fabric.
Either SAN Volume Controller data migration scenario requires scheduled downtime, to
connect the application server to the SAN Volume Controller. Depending on the type of server
used, the system might need to be shut down, the original LUNs removed and the new LUNs
discovered. The specific steps for moving the servers are addressed in subsequent sections.
7.4.2 Connect SAN Volume Controller to the fabric for migration

Start by configuring the zone set for the new target storage unit and the SAN Volume
Controller. There are two possibilities for this zoning. Table 7-3 shows the zones created in
each case. Zone the target SAN Volume Controller cluster to the target storage, and the
source SAN Volume Controller cluster to the source storage.
If the migration uses Metro Mirror, there are a source and target SAN Volume Controller
cluster. Two additional zones must be created for the application server and zoned to the
source SAN Volume Controller and the target storage.
Table 7-3 Zone requirements for migration

Migration using Metro Mirror Migration using volume migration
Source SAN Volume Controller and source SAN Volume Controller and source storage
storage
Target SAN Volume Controller and target storage SAN Volume Controller and target storage

Migration using Metro Mirror Migration using volume migration
Application server and source SAN Volume Application server and SAN Volume Controller
Controller
Application server and target SAN Volume Application server and target SAN Volume
Controller Controller
Source and target SAN Volume Controller (for

Metro Mirror paths)
If the SAN Volume Controller will remain after migration, additional zones are required for the
application server and the target SAN Volume Controller:
򐂰 For volume migration, three new zones are required to connect the SAN Volume Controller
to both source and target storage. These zones also attach the application server to the
new target storage.
򐂰 For migration using Metro Mirror, a zone for SAN Volume Controller and the application
server are created instead of the application server and target storage.
7.4.3 Remove SAN Volume Controller from the fabric after migration
The SAN Volume Controller can remain in the fabric for continued storage virtualization if
wanted. To remove the SAN Volume Controller, perform the following steps:
1. Shut down the server.
1. Remove the SAN Volume Controller and application server zone.
2. Create and activate a zone for the application server and the new storage.
3. Remove the SAN Volume Controller zones to the storage.
4. Bring the application server up on the new storage.
5. Disconnect and remove the SAN Volume Controller.
Remember: Additional steps on the storage are required to map the application server to
the target volumes.
7.4.4 Back-End storage consideration

When planning the configuration of external storage systems for use with SAN Volume
Controller systems, consider the following details:
򐂰 All SAN Volume Controller nodes in a system must be able to connect to the same set of
storage system ports on each device. A system that contains nodes that cannot connect to
the same set of storage-system ports is considered degraded. In this situation, a system
error is logged that requires a repair action. This action is important on storage systems
that have exclusion rules that determine to which host bus adapter (HBA) worldwide node
names (WWNNs) a storage partition can be mapped.
򐂰 A storage-system logical unit (LU) must not be shared between the SAN Volume
Controller and a host. You can configure certain storage systems to safely share
resources between the SAN Volume Controller system and direct-attached hosts. This
type of configuration is described as a split storage system. You must configure the
storage system and SAN so the SAN Volume Controller system cannot access LUs that
any other host or SAN Volume Controller system can access.

򐂰 This split storage system configuration can be arranged by storage system LUN mapping
and masking. If the split storage system configuration is not guaranteed, data corruption
can occur.
򐂰 The SAN Volume Controller supports configurations where a storage system is split
between two SAN Volume Controllers. You must configure the storage system and SAN so
that the SAN Volume Controller system cannot access LUs that any other host or SAN
Volume Controller system can access. You can use storage system LUN mapping and
masking to arrange for this configuration. If this configuration is not guaranteed, data
corruption can occur.
Attention: Avoid configuring a storage system to present the same LU to more than one
SAN Volume Controller system. This configuration is not supported and is likely to cause
undetected data loss or corruption.
For the latest support information, see the SAN Volume Controller website at:
7.4.5 Unsupported storage systems

When a storage system is detected on the SAN, the SAN Volume Controller attempts to
recognize it using its inquiry data. If the device is not supported, the SAN Volume Controller
configures the device as a generic device. A generic device might not function correctly when
it is addressed by a SAN Volume Controller system, especially under failure scenarios.
However, the SAN Volume Controller system does not regard accessing a generic device as
an error condition and does not log an error. Managed disks (MDisks) that are presented by
generic devices are not eligible to be used as quorum disks.
7.4.6 Host attachment

Host attachment consideration must be taken seriously from compatibility perspective. OS
host levels must satisfy latest OS patch levels and SAN compatibility requirements, and be
compatible with SAN Volume Controller code level.
The SAN Volume Controller supports IBM and non-IBM hosts. Check the interoperability
information at the following web address:
7.5 Migrating SAN disks to SAN Volume Controller volumes

and back to SAN
This section addresses how to move two LUNs from an AIX server connected to a storage
subsystem through SAN over to the SAN Volume Controller.
Manage the LUNs with the SAN Volume Controller. Move them between other managed
disks, and then back to image mode disks so that they can be masked and mapped back to
the AIX server directly.

Using this procedure allows you to perform any of the following activities in your environment:
򐂰 Move the SAN LUNs of an AIX server and virtualize those LUNs through the SAN Volume
Controller. Perform these steps when introducing the SAN Volume Controller into your
environment.
Your host downtime is only a few minutes while you remap and remask disks using your
storage subsystem LUN management tool.
򐂰 Move data between storage subsystems while your AIX server is still running and
servicing your business application.
Perform this activity if you are removing a storage subsystem from your SAN environment
and want to move the data onto more appropriate LUNs. When selecting LUNs, take into
account availability, performance, and redundancy.
򐂰 Move the LUNs on your AIX server back to image mode volumes so that they can be
remapped and remasked directly to the AIX server.
Figure 7-8 shows our AIX server connected to our SAN infrastructure. It has two LUNs
(hdisk1 and hdisk2) that are zoned directly to it from our storage subsystem. In this example,
a heavy I/O load was simulated to see how the migration process affects the host side.
Zoning for migration scenarios
AIX
Host
SAN
Green Zone
IBM or OEM
Storage
Subsystem
Figure 7-8 AIX SAN environment

The hdisk1 disk makes up the itsovg01 LVM group, and the hdisk2 disk makes up the
itsovg02 LVM group, as shown in Example 7-1.
Example 7-1 AIX SAN configuration

# lspv
hdisk0 00ce0c7b1fcfc0e5 rootvg active
hdisk1 00ce0c7b69d8ea93 itsovg01 active
hdisk2 00ce0c7b69d9454b itsovg02 active
#
# lsdev -Cc disk
hdisk0 Available Virtual SCSI Disk Drive
hdisk1 Available 02-08-02 MPIO 2810 XIV Disk
hdisk2 Available 02-08-02 MPIO 2810 XIV Disk
#
# lscfg |grep -i disk
* hdisk0 U9119.590.83E0C7B-V101-C2-T1-L8100000000000000 Virtual SCSI Disk Drive
* hdisk1 U5791.001.99B0832-P2-C04-T1-W50017380102F0180-L1000000000000 MPIO 2810
XIV Disk
* hdisk2 U5791.001.99B0832-P2-C04-T1-W50017380102F0180-L2000000000000 MPIO 2810
XIV Disk
#
The AIX server represents a typical SAN environment. The host directly uses LUNs that were
created on a SAN storage subsystem (Figure 7-8 on page 204):
򐂰 The HBA cards on the AIX server are zoned so that they are in the Green Zone (dotted
line) with the storage subsystem.
򐂰 The two LUNs, hdisk1 and hdisk2, are defined on the storage subsystem. They are
directly zoned and available to the AIX server.
7.5.1 Connecting the SAN Volume Controller to your SAN fabric

This section addresses the steps to take to introduce SAN Volume Controller into your SAN
environment. You can accomplish this task without any downtime to any host or application
that also uses your storage area network. If you have an SAN Volume Controller already
connected, skip to 7.5.2, “Preparing your SAN Volume Controller to virtualize disks” on
page 206.
To connect the SAN Volume Controller to your SAN fabric, perform the following tasks:
1. Assemble your SAN Volume Controller components (nodes, UPS, and SSPC Console if
available).
2. Cable the SAN Volume Controller correctly.
3. Power the SAN Volume Controller on.
4. Verify that the SAN Volume Controller is visible on your SAN.
5. Create and configure your SAN Volume Controller cluster.
6. Create these additional zones:
– An SAN Volume Controller node zone (Black Zone)
– A storage zone (Red Zone)
– A zone for every host initiator (HBA) (Blue Zone)

Figure 7-9 shows the environment and defined zones.
AIX
Host
SVC
I/O grp0
SVC
SVC
SAN
Green Zone
IBM or OEM IBM or OEM
Storage Storage Red Zone
Subsystem Subsystem
Blue Zone
Black Zone
Figure 7-9 SAN environment with SAN Volume Controller attached
7.5.2 Preparing your SAN Volume Controller to virtualize disks

This section describes the preparatory tasks you must perform before taking your AIX server
offline. These tasks are all nondisruptive activities and do not affect your SAN fabric or your
existing SAN Volume Controller configuration.
Creating a storage pool

When moving the two AIX LUNs to the SAN Volume Controller, they are in image mode, so
create a storage pool to hold them. In this example, the storage pool to hold the LUNs is
named aix_imgmdg.
Create an empty storage pool for these disks using the commands shown in Example 7-2.
Example 7-2 Create empty Storage pool

IBM_2145:ITSO1:superuser>svctask mkmdiskgrp -name aix_imgmdg -ext 512
MDisk Group, id [2], successfully created
IBM_2145:ITSO1:superuser>
IBM_2145:ITSO1:superuser>lsmdiskgrp
id name status mdisk_count vdisk_count capacity extent_size free_capacity
virtual_capacity used_capacity real_capacity overallocation warning easy_tier
easy_tier_status

2 aix_imgmdg online 0 0 0 512 0 0.00MB 0.00MB 0.00MB 0 0 auto inactive
Creating host definition

If you prepare the zones correctly, the SAN Volume Controller can see the HBA adapters for
the AIX server on the fabric.
First, get the worldwide name (WWN) for the HBA of your AIX server. Make sure that you
have the correct WWN to create a zone for every HBA in the Blue Zone and to reduce the AIX
server downtime.
Example 7-3 shows the commands to get the WWN. In this example, the host has WWNs of
10000000C95ADE4D and 10000000C9648394.
Example 7-3 Discover your WWN

# lsdev -Cc adapter|grep fcs
fcs0 Available 02-08 FC Adapter
fcs1 Available 03-08 FC Adapter
#
# lscfg -vpl fcs0
fcs0 U5791.001.99B0832-P2-C04-T1 FC Adapter
Part Number.................03N5014
EC Level....................A
Serial Number...............1D719080D1
Manufacturer................001D
Customer Card ID Number.....280D
FRU Number.................. 03N5014
Device Specific.(ZM)........3
Network Address.............10000000C95ADE4D
ROS Level and ID............02C82138
Device Specific.(Z0)........1036406D
Device Specific.(Z1)........00000000
Device Specific.(Z4)........FFC01159
Device Specific.(Z5)........02C82138
Device Specific.(Z8)........20000000C95ADE4D
Device Specific.(Z9)........BS2.10X8
Device Specific.(ZA)........B1D2.10X8
Device Specific.(ZB)........B2D2.10X8
Device Specific.(ZC)........00000000
Hardware Location Code......U5791.001.99B0832-P2-C04-T1
PLATFORM SPECIFIC
Name: fibre-channel
Model: LP11000
Node: fibre-channel@1
Device Type: fcp
Physical Location: U5791.001.99B0832-P2-C04-T1

#
# lscfg -vpl fcs1
fcs1 U5791.001.99B082W-P1-C04-T1 FC Adapter
Part Number.................10N8620
Serial Number...............1B71704A6C
Manufacturer................001B
EC Level....................A
Customer Card ID Number.....5759
FRU Number.................. 10N8620
Device Specific.(ZM)........3
Network Address.............10000000C9648394
ROS Level and ID............02C82138
Device Specific.(Z0)........1036406D
Device Specific.(Z4)........FFC01159
Device Specific.(Z9)........BS2.10X8
Device Specific.(ZA)........B1F2.10X8
Device Specific.(ZB)........B2F2.10X8
Device Specific.(ZC)........00000000
Hardware Location Code......U5791.001.99B082W-P1-C04-T1
PLATFORM SPECIFIC
Name: fibre-channel
Model: LP11000
Node: fibre-channel@1
Device Type: fcp
Physical Location: U5791.001.99B082W-P1-C04-T1
#
The svcinfo lshbaportcandidate command lists all of the WWNs that are not yet allocated
to a host on the SAN fabric. Example 7-4 shows the output of the nodes that it found in the
example SAN fabric. If the port does not show up, there is a zone configuration problem.
Example 7-4 Add the host to the SAN Volume Controller
IBM_2145:ITSO-CLS2:admin>svcinfo lshbaportcandidate
id
10000000C95ADE4D
10000000C9648394
IBM_2145:ITSO-CLS2:admin>

After verifying that the SAN Volume Controller can see the host ITSO_AIX, create the host
entry and assign the WWN to this entry (Example 7-5).
Example 7-5 Create the host entry

IBM_2145:ITSO-CLS2:admin>svctask mkhost -name ITSO_AIX -hbawwpn
10000000C95ADE4D:10000000C9648394
Host, id [0], successfully created
IBM_2145:ITSO-CLS2:admin>svcinfo lshost ITSO_AIX
id 0
name ITSO_AIX
port_count 2
type generic
mask 1111
iogrp_count 4
WWPN 10000000C9648394
node_logged_in_count 2
state inactive
WWPN 10000000C95ADE4D
node_logged_in_count 2
state inactive
Verifying that you can see your storage subsystem

Display the storage subsystem in the SAN Volume Controller with the svcinfo lscontroller
command (Example 7-6). In this example, XIV is used as controller1 with id 1, and will be
migrated to DS8000 as a controller with id 0.
Example 7-6 Discover the storage controller

IBM_2145:ITSO1:superuser>lscontroller
id controller_name ctrl_s/n vendor_id product_id_low
product_id_high
0 controller0 75ABTV2FFFF IBM 2107900
1 controller1 102F0000 IBM 2810XIV-
LUN-0
Tip: The svctask chcontroller command allows you to change the discovered storage
subsystem name in SAN Volume Controller. In complex SANs, change your storage
subsystem to a more meaningful name.
Getting the disk serial numbers

To avoid creating the wrong volumes when there are many available unmanaged MDisks that
are seen by the SAN Volume Controller, locate the following identifiers:
򐂰 The LUN unique_id from the AIX side
򐂰 The WWN tag from the storage subsystem
In the example, the WWN comes from an XIV disk subsystem.
When you discover these MDisks, confirm that you have the correct serial numbers before
creating the image mode volumes.

From the XIV system, use XCLI to obtain the WWN LUN as shown in Example 7-7.
Example 7-7 Use XIV XCLI to obtain WWN for LUN attached to ITSO_AIX host
XIV_7804143>>mapping_list host=ITSO_AIX
LUN Volume Size Master Serial Number Locked
6 lpar3_1 17 14276 no
7 lpar3_2 17 14277 no
XIV_7804143>>
XIV_7804143>>vol_list vol=lpar3_1 -t "wwn"

WWN
00173800102F37C4
XIV_PFE3_7804143>>vol_list vol=lpar3_2 -t "wwn"
WWN
00173800102F37C5
XIV_PFE3_7804143>>
Obtain the hdisk unique_id from ITSO_AIX as shown in Example 7-8.
Example 7-8 Obtain hdisk unique_id from ITSO_AIX

# lsattr -El hdisk1 | grep unique_id
unique_id 2611200173800102F37C4072810XIV03IBMfcp
# lsattr -El hdisk2 | grep unique_id

unique_id 2611200173800102F37C5072810XIV03IBMfcp
#
7.5.3 Moving the LUNs to the SAN Volume Controller

Next, move the LUNs that are assigned to the AIX server and reassign them to the SAN
Volume Controller.
Because you want to move only the LUN that holds your application and data files, move that
LUN without rebooting the host. You must unmount only the file system and vary off the
volume group (VG) to ensure data integrity after the reassignment.
Important: Moving LUNs to the SAN Volume Controller requires that the Subsystem
Device Driver is installed on the AIX server. You can install the Subsystem Device Driver
ahead of time. However, it might require an outage of your host to do so. The latest driver
information is available at following web address:
https://www-304.ibm.com/support/docview.wss?uid=ssg1S7001350
Subsystem Device Driver can coexist with other multipath drivers only during migration.
To move both LUNs at the same time, perform the following steps:
1. Confirm that the Subsystem Device Driver is installed.
2. Unmount the file system and vary off the VGs:
a. Stop the applications that are using the LUNs.
b. Unmount those file systems with the unmount MOUNT_POINT command.

c. If the file systems are part of a Logical Volume Manager (LVM) volume, deactivate that
VG with the varyoffvg VOLUMEGROUP_NAME command.
Example 7-9 shows the commands run in the example on ITSO_AIX.
Example 7-9 AIX command sequence

# lslpp -l |grep -i sdd
devices.sddpcm.61.rte 2.6.0.3 COMMITTED IBM SDD PCM for AIX V61
devices.sddpcm.61.rte 2.6.0.3 COMMITTED IBM SDD PCM for AIX V61
# lsfs
Name Nodename Mount Pt VFS Size Options Auto
Accounting
/dev/hd4 -- / jfs2 1048576 -- yes no
/dev/hd1 -- /home jfs2 6291456 -- yes no
/dev/hd2 -- /usr jfs2 6291456 -- yes no
/dev/hd9var -- /var jfs2 3080192 -- yes no
/dev/hd3 -- /tmp jfs2 524288 -- yes no
/dev/hd11admin -- /admin jfs2 262144 -- yes no
/proc -- /proc procfs -- -- yes
no
/dev/hd10opt -- /opt jfs2 2097152 -- yes no
/dev/livedump -- /var/adm/ras/livedump jfs2 524288 -- yes no
/dev/odm -- /dev/odm vxodm -- -- no no
/dev/fslv01 -- /itsofs01 jfs2 33423360 rw yes no
/dev/fslv02 -- /itsofs02 jfs2 33423360 rw yes no
# lsvg -o
itsovg02
itsovg01
rootvg
# lsvg -l itsovg01
itsovg01:
LV NAME TYPE LPs PPs PVs LV STATE MOUNT POINT
loglv01 jfs2log 1 1 1 open/syncd N/A
fslv01 jfs2 510 510 1 open/syncd /itsofs01
# lsvg -l itsovg02
itsovg02:
fslv02 jfs2 510 510 1 open/syncd /itsofs02
# umount /itsofs01
# umount /itsofs02
# varyoffvg itsovg01
# varyoffvg itsovg02
3. From the storage side (the XIV system in the example), unmap the disks from the
ITSO_AIX server and remap the disks to the SAN Volume Controller.

4. From the SAN Volume Controller, discover the new disks with the svctask detectmdisk
command. The disks are discovered and named mdiskN, where N is the next available
mdisk number (starting from 0). Example 7-10 shows the commands used in the example
to discover the MDisks and verify that you have the correct MDisks. The bolded part
shows the LUNs from the XIV system and their unique IDs.
Example 7-10 Discover the new MDisks

IBM_2145:ITSO1:superuser>detectmdisk
IBM_2145:ITSO1:superuser>lsmdisk
id name status mode mdisk_grp_id mdisk_grp_name capacity ctrl_LUN_#
controller_name UID tier
8 mdisk8 online unmanaged 16.0GB
0000000000000005 controller1
00173800102f37c4000000000000000000000000000000000000000000000000 generic_hdd
0000000000000006 controller1
00173800102f37c5000000000000000000000000000000000000000000000000 generic_hdd
Important: Match your discovered MDisk serial numbers (UID in the svcinfo lsmdisk
command task display) with the serial number that you discovered earlier.
5. After you verify that you have the correct MDisks, rename them to avoid confusion in the
future (Example 7-11).
Example 7-11 Rename the MDisks

IBM_2145:ITSO1:superuser>svctask chmdisk -name ITSO_AIX01 mdisk8
IBM_2145:ITSO1:superuser>svctask chmdisk -name ITSO_AIX02 mdisk9
id name status mode mdisk_grp_id mdisk_grp_name capacity
ctrl_LUN_# controller_name UID
tier
8 ITSO_AIX01 online unmanaged 16.0GB
0000000000000005 controller1
00173800102f37c4000000000000000000000000000000000000000000000000 generic_hdd
0000000000000006 controller1
00173800102f37c5000000000000000000000000000000000000000000000000 generic_hdd
6. Create the image mode volumes with the svctask mkvdisk command and the option
-vtype image (Example 7-12). This command virtualizes the disks in the same layout as
though they were not virtualized.
Example 7-12 Create the image mode volumes

IBM_2145:ITSO1:superuser>mkvdisk -mdiskgrp aix_imgmdg -iogrp 0 -vtype image -mdisk
ITSO_AIX01 -name itso_aix_hdisk1
Virtual Disk, id [2], successfully created
IBM_2145:ITSO1:superuser>mkvdisk -mdiskgrp aix_imgmdg -iogrp 0 -vtype image -mdisk

ITSO_AIX02 -name itso_aix_hdisk2
Virtual Disk, id [3], successfully created

7. Map the new image mode volumes to the host (Example 7-13).
Example 7-13 Map the volumes to the host

IBM_2145:ITSO1:superuser>svctask mkvdiskhostmap -host ITSO_AIX itso_aix_hdisk1
Virtual Disk to Host map, id [2], successfully created
IBM_2145:ITSO1:superuser>svctask mkvdiskhostmap -host ITSO_AIX itso_aix_hdisk2

Virtual Disk to Host map, id [3], successfully created
Tip: While the application is in a quiescent state, you can use FlashCopy to copy the new
image volumes onto other volumes. You do not need to wait until the FlashCopy process is
completed before starting your application.
To put the image mode volumes online, perform the following steps:
1. Remove the old disk definitions using the rmdev -dl <hdisk> command.
2. Run the cfgmgr -vs command to rediscover the available LUNs.
3. If your application and data are on an LVM volume, rediscover the VG, and then run the
varyonvg VOLUME_GROUP command to activate the VG.
4. Mount your file systems with the mount /MOUNT_POINT command.
You are now ready to start your application.
7.5.4 Migrating image mode volumes to volumes

While the AIX server is running and the file systems are in use, migrate the image mode
volumes onto striped volumes. The extents are spread over three other MDisks.
Preparing MDisks for striped mode volumes

From the new storage subsystem, perform these tasks:
򐂰 Create and allocate three LUNs from your new storage subsystem to the SAN Volume
Controller.
򐂰 Discover the LUNs as MDisks.
򐂰 Rename these LUNs to more meaningful names.
򐂰 Create a storage pool.
򐂰 Put all these MDisks into this storage pool.
The output of these commands in the example is shown in Example 7-14.
Example 7-14 Creating a storage pool

IBM_2145:ITSO1:superuser>mkmdiskgrp -name itso_aix_new -ext 512
tier

4084400500000000 controller0
6005076303ffce63000000000000840500000000000000000000000000000000 generic_hdd
4085400400000000 controller0
6005076303ffce63000000000000850400000000000000000000000000000000 generic_hdd
4085400500000000 controller0
6005076303ffce63000000000000850500000000000000000000000000000000 generic_hdd
8 ITSO_AIX01 online image 2 aix_imgmdg 16.0GB
0000000000000005 controller1
00173800102f37c4000000000000000000000000000000000000000000000000 generic_hdd
0000000000000006 controller1
00173800102f37c5000000000000000000000000000000000000000000000000 generic_hdd
IBM_2145:ITSO1:superuser>chmdisk -name itso_aix_md1 mdisk5
IBM_2145:ITSO1:superuser>addmdisk -mdisk itso_aix_md1 itso_aix_new
tier
5 itso_aix_md1 online managed 1 itso_aix_new 15.0GB
4084400500000000 controller0
6005076303ffce63000000000000840500000000000000000000000000000000 generic_hdd
4085400400000000 controller0
6005076303ffce63000000000000850400000000000000000000000000000000 generic_hdd
4085400500000000 controller0
6005076303ffce63000000000000850500000000000000000000000000000000 generic_hdd
0000000000000005 controller1
00173800102f37c4000000000000000000000000000000000000000000000000 generic_hdd
0000000000000006 controller1
00173800102f37c5000000000000000000000000000000000000000000000000 generic_hdd
Migrating the volumes

Migrate the image mode volumes onto striped volumes with the svctask migratevdisk
command (Example 7-15 on page 215). While the migration is running, the AIX server is still
running and you can still access the files.
Before starting the migration process, run heavy I/O load to hdisk1 and hdisk2 on the AIX side
and collect performance data using the nmon tool.
To check the overall progress of the migration, use the svcinfo lsmigrate command as
shown in Example 7-15 on page 215. Listing the storage pool with the svcinfo lsmdiskgrp

command shows that the free capacity on the old storage pool is slowly increasing. Capacity
is being freed as those extents are moved to the new storage pool.
Use CLI command svctask lsvdiskextent to track the progress as extents move from image
mode volumes to new striped volumes.
Example 7-15 Migrating image mode volumes to striped volumes

IBM_2145:ITSO1:superuser>lsvdiskextent 2
id number_extents
0 32
id number_extents
0 32
IBM_2145:ITSO1:superuser>migratevdisk -vdisk itso_aix_hdisk1 -mdiskgrp

itso_aix_new
IBM_2145:ITSO1:superuser>migratevdisk -vdisk itso_aix_hdisk2 -mdiskgrp
itso_aix_new
IBM_2145:ITSO1:superuser>lsmigrate
migrate_type MDisk_Group_Migration
progress 0
migrate_source_vdisk_index 3
migrate_target_mdisk_grp 1
max_thread_count 4
migrate_source_vdisk_copy_id 0
progress 3
max_thread_count 4
id number_extents
5 3
6 3
7 3
8 23
id number_extents
5 4
6 2
7 3
9 23
progress 62
max_thread_count 4

progress 65
max_thread_count 4
id number_extents
5 11
6 11
7 10
id number_extents
5 11
6 11
7 10
After this task is completed, Example 7-16 shows that the volumes are spread over three
MDisks in the itso_aix_new storage pool. The old storage pool is empty.
Example 7-16 Migration complete

IBM_2145:ITSO1:superuser>svcinfo lsmdiskgrp itso_aix_new
id 1
name itso_aix_new
status online
mdisk_count 3
vdisk_count 2
capacity 43.50GB
extent_size 512
free_capacity 11.50GB
virtual_capacity 32.00GB
used_capacity 32.00GB
real_capacity 32.00GB
overallocation 73
warning 0
easy_tier auto
easy_tier_status inactive
tier generic_ssd
tier_mdisk_count 0
tier_capacity 0.00MB
tier_free_capacity 0.00MB
tier generic_hdd
tier_mdisk_count 3
tier_capacity 43.50GB
tier_free_capacity 11.50GB
IBM_2145:ITSO1:superuser>svcinfo lsmdiskgrp aix_imgmdg

id 2
name aix_imgmdg

status online
mdisk_count 2
vdisk_count 0
capacity 32.00GB
extent_size 512
free_capacity 32.00GB
virtual_capacity 0.00MB
used_capacity 0.00MB
real_capacity 0.00MB
overallocation 0
warning 0
easy_tier auto
easy_tier_status inactive
tier generic_ssd
tier_mdisk_count 0
tier_capacity 0.00MB
tier_free_capacity 0.00MB
tier generic_hdd
tier_mdisk_count 2
tier_capacity 32.00GB
tier_free_capacity 32.00GB
IBM_2145:ITSO1:superuser>lsvdisk
id name IO_group_id IO_group_name status mdisk_grp_id mdisk_grp_name
capacity type FC_id FC_name RC_id RC_name vdisk_UID
fc_map_count copy_count fast_write_state se_copy_count
2 itso_aix_hdisk1 0 io_grp0 online 1 itso_aix_new
16.00GB striped 60050768018106209800000000000009 0
1 not_empty 0
3 itso_aix_hdisk2 0 io_grp0 online 1 itso_aix_new

16.00GB striped 6005076801810620980000000000000A 0
1 not_empty 0
The migration to the SAN Volume Controller is complete. You can remove the original MDisks
from the SAN Volume Controller, and the LUNs from the storage subsystem.
If the LUNs are the ones used last on your storage subsystem, remove them from your SAN
fabric using the following steps, shown in Example 7-17:
1. Remove MDisks from the storage pool.
2. Remove the storage pool.
3. Delete the storage pool.
4. Remove mapping from SAN Volume Controller on your back-end storage.
5. Run the detectmdisk command on SAN Volume Controller to be sure that MDisks are
gone.
Example 7-17 Using SAN Volume Controller to remove the storage pool
easy_tier_status

1 itso_aix_new online 3 2 43.50GB 512 11.50GB
32.00GB 32.00GB 32.00GB 73 0 auto
inactive
2 aix_imgmdg online 2 0 32.00GB 512 32.00GB
0.00MB 0.00MB 0.00MB 0 0 auto
inactive
tier
4084400500000000 controller0
6005076303ffce63000000000000840500000000000000000000000000000000 generic_hdd
4085400400000000 controller0
6005076303ffce63000000000000850400000000000000000000000000000000 generic_hdd
4085400500000000 controller0
6005076303ffce63000000000000850500000000000000000000000000000000 generic_hdd
8 ITSO_AIX01 online managed 2 aix_imgmdg 16.0GB
0000000000000005 controller1
00173800102f37c4000000000000000000000000000000000000000000000000 generic_hdd
9 ITSO_AIX02 online managed 2 aix_imgmdg 16.0GB
0000000000000006 controller1
00173800102f37c5000000000000000000000000000000000000000000000000 generic_hdd
IBM_2145:ITSO1:superuser>rmmdisk -mdisk ITSO_AIX01 aix_imgmdg

IBM_2145:ITSO1:superuser>rmmdisk -mdisk ITSO_AIX02 aix_imgmdg
easy_tier_status
32.00GB 32.00GB 32.00GB 73 0 auto
inactive
2 aix_imgmdg online 0 0 0 512 0
0.00MB 0.00MB 0.00MB 0 0 auto
inactive
tier
4084400500000000 controller0
6005076303ffce63000000000000840500000000000000000000000000000000 generic_hdd
4085400400000000 controller0
6005076303ffce63000000000000850400000000000000000000000000000000 generic_hdd
4085400500000000 controller0
6005076303ffce63000000000000850500000000000000000000000000000000 generic_hdd

0000000000000005 controller1
00173800102f37c4000000000000000000000000000000000000000000000000 generic_hdd
0000000000000006 controller1
00173800102f37c5000000000000000000000000000000000000000000000000 generic_hdd
IBM_2145:ITSO1:superuser>rmmdiskgrp aix_imgmdg
easy_tier_status
32.00GB 32.00GB 32.00GB 73 0 auto
inactive
tier
4084400500000000 controller0
6005076303ffce63000000000000840500000000000000000000000000000000 generic_hdd
4085400400000000 controller0
6005076303ffce63000000000000850400000000000000000000000000000000 generic_hdd
4085400500000000 controller0
6005076303ffce63000000000000850500000000000000000000000000000000 generic_hdd
7.5.5 Performance analysis

As shown in Figure 7-10 on page 220, there was no performance impact to host I/O during
the test. During the migration, there was higher oscillation in the I/O profile, but SAN Volume
Controller sustained I/O requests from the host side.
On the graph shows three segments:

򐂰 Before migration
򐂰 During migration
򐂰 Post migration
With the I/O pattern constant during the test, I/O activity increased after moving from image
mode MDisks (not striped) to stripe MDisks. Therefore, always use striping to improve
performance.
I/O Pattern: BS=4k, Mixed R/W 50/50, random, 50% Cache Hit

Figure 7-10 ITSO_AIX System summary
Figure 7-11 shows the disk transfers per second during the example migration.
Figure 7-11 ITSO_AIX disk IOPS
7.5.6 Preparing to migrate from the SAN Volume Controller

Before changing the AIX server LUNs from being accessed by the SAN Volume Controller as
volumes to being directly accessed, convert the volumes into image mode volumes.
This activity might be needed for one of these reasons:

򐂰 You purchased a new storage subsystem, and are using the SAN Volume Controller as a
tool to migrate the new storage system.
򐂰 You used the SAN Volume Controller to FlashCopy or Metro Mirror a volume onto another
volume, and no longer need that host connected to the controller.
򐂰 You want to move a host and its data currently connected to the SAN Volume Controller to
a site where there is no SAN Volume Controller.
򐂰 Changes to your environment no longer require this host to use the SAN Volume
Controller.
There are other preparatory activities to be performed before we shut down the host and
reconfigure the zoning and LUN mapping. This section covers those activities.

If you are moving the data to a new storage subsystem, the subsystem must be connected to
your SAN fabric, powered on, and visible from your SAN switches. Your environment should
look similar to the environment shown in Figure 7-12.
AIX
Host
SVC
I/O grp0
SVC
SVC
SAN
Green Zone
IBM or OEM IBM or OEM
Storage Storage Red Zone
Subsystem Subsystem
Blue Zone
Black Zone
Figure 7-12 Environment with SAN Volume Controller
To prepare for the migration, perform the following steps:

򐂰 Set up the SAN configuration so that all of the zones are created.
򐂰 Add the new storage subsystem to the Red Zone so that the SAN Volume Controller can
communicate with it directly.
򐂰 Create a Green Zone for the host to use when you are ready for it to directly access the
disk. This step must take place after the host is removed from the SAN Volume Controller.
After your zone configuration is set up correctly, use the svcinfo lscontroller command to
force SAN Volume Controller to recognize the new storage subsystem controller
(Example 7-18). In addition, you might want to rename the controller to a more meaningful
name using the svctask chcontroller -name command.
Example 7-18 Discovering the new storage subsystem

IBM_2145:ITSO1:superuser>lscontroller
id controller_name ctrl_s/n vendor_id product_id_low
product_id_high
0 controller0 75ABTV2FFFF IBM 2107900
1 controller1 102F0000 IBM 2810XIV- LUN-0

7.5.7 Creating new LUNs
The example storage subsystem has two LUNs that are masked so the SAN Volume
Controller can see them. These LUNs link directly to the host, removing the volumes that it
currently has. To check that the SAN Volume Controller can use the LUNs, issue the svctask
detectmdisk command as shown in Example 7-19.
In the example, two 16 GB LUNs are on the XIV subsystem. Migrate them back to image
mode volumes and the XIV subsystem in one step.
Example 7-19 Discover the new MDisks

4084400500000000 controller0
6005076303ffce63000000000000840500000000000000000000000000000000 generic_hdd
4085400400000000 controller0
6005076303ffce63000000000000850400000000000000000000000000000000 generic_hdd
4085400500000000 controller0
6005076303ffce63000000000000850500000000000000000000000000000000 generic_hdd
0000000000000005 controller1
00173800102f37c4000000000000000000000000000000000000000000000000 generic_hdd
0000000000000006 controller1
00173800102f37c5000000000000000000000000000000000000000000000000 generic_hdd
Although the MDisks do not stay in the SAN Volume Controller for long, rename them.
Change the names to more meaningful ones so that you do not confuse them with other
MDisks. Also, create the storage pools to hold your new MDisks as shown in Example 7-20.
Example 7-20 Rename the MDisks

IBM_2145:ITSO1:superuser>svctask chmdisk -name AIX_MIG01 mdisk5
IBM_2145:ITSO1:superuser>svctask chmdisk -name AIX_MIG02 mdisk6
IBM_2145:ITSO1:superuser>svctask mkmdiskgrp -name ITSO_AIXMIG -ext 512
easy_tier_status
32.00GB 32.00GB 32.00GB 73 0 auto
inactive
2 ITSO_AIXMIG online 0 0 0 512 0
0.00MB 0.00MB 0.00MB 0 0 auto
inactive
Your SAN Volume Controller environment is now ready for the volume migration to image
mode volumes.

7.5.8 Migrating the managed volumes
While the AIX server is still running, migrate the managed volumes onto the new MDisks
using image mode volumes. Perform this action with the svctask migratetoimage command
as shown in Example 7-21.
Example 7-21 Migrate the volumes to image mode volumes

IBM_2145:ITSO1:superuser>migratetoimage -vdisk itso_aix_hdisk1 -mdisk AIX_MIG01
-mdiskgrp ITSO_AIXMIG
tier
4084400500000000 controller0
6005076303ffce63000000000000840500000000000000000000000000000000 generic_hdd
4085400400000000 controller0
6005076303ffce63000000000000850400000000000000000000000000000000 generic_hdd
4085400500000000 controller0
6005076303ffce63000000000000850500000000000000000000000000000000 generic_hdd
8 AIX_MIG01 online image 2 ITSO_AIXMIG 16.0GB
0000000000000005 controller1
00173800102f37c4000000000000000000000000000000000000000000000000 generic_hdd
0000000000000006 controller1
00173800102f37c5000000000000000000000000000000000000000000000000 generic_hdd
migrate_type Migrate_to_Image
progress 25
migrate_target_mdisk_index 8
max_thread_count 4
progress 3
max_thread_count 4
id number_extents
5 4
6 5
7 3
8 20

id number_extents
5 6
6 7
7 6
9 13
id number_extents
8 32
id number_extents
9 32
IBM_2145:ITSO1:superuser>IBM_2145:ITSO1:superuser>migratetoimage -vdisk
itso_aix_hdisk1 -mdisk AIX_MIG01 -mdiskgrp ITSO_AIXMIG
tier
4084400500000000 controller0
6005076303ffce63000000000000840500000000000000000000000000000000 generic_hdd
4085400400000000 controller0
6005076303ffce63000000000000850400000000000000000000000000000000 generic_hdd
4085400500000000 controller0
6005076303ffce63000000000000850500000000000000000000000000000000 generic_hdd
0000000000000005 controller1
00173800102f37c4000000000000000000000000000000000000000000000000 generic_hdd
0000000000000006 controller1
00173800102f37c5000000000000000000000000000000000000000000000000 generic_hdd
progress 25
max_thread_count 4
progress 3

max_thread_count 4
id number_extents
5 4
6 5
7 3
8 20
id number_extents
5 6
6 7
7 6
9 13
id number_extents
8 32
id number_extents
9 32
During the migration, the AIX server is unaware that its data is being physically moved
between storage subsystems.
After the migration is complete, the image mode volumes are ready to be removed from the
AIX server. The real LUNs can be mapped and masked directly to the host by using the
storage subsystems tool.
7.5.9 Removing the LUNs from the SAN Volume Controller

The next step requires a downtime while you remap and remask the disks so the host sees
them directly through the Green Zone. Because your LUNs hold only data files and you are
using a unique VG, you can remap and remask the disks without rebooting the host. The only
requirement is that you unmount the file system and vary off the VG to ensure data integrity
after the reassignment.
Remember: Moving LUNs to another storage system might need a driver other than SDD.
Check with the storage subsystems vendor to see which driver you need. You might be
able to install this driver ahead of time.
Follow these steps to remove the SAN Volume Controller:

1. Confirm that the correct device driver for the new storage subsystem is loaded. In the
example, you have already installed the XIV host connection kit for AIX.
2. Shut down any applications and unmount the file systems:
a. Stop the applications that are using the LUNs.

b. Unmount those file systems with the umount MOUNT_POINT command.
c. If the file systems are an LVM volume, deactivate that VG with the varyoffvg
VOLUMEGROUP_NAME command.
3. Remove the volumes from the host by using the svctask rmvdiskhostmap command
(Example 7-22). To double-check that the volumes are removed, use the svcinfo
lshostvdiskmap command. The command shows that these disks are no longer mapped
to the AIX server.
Example 7-22 Remove the volumes from the host

IBM_2145:ITSO1:superuser>rmvdiskhostmap -host ITSO_AIX itso_aix_hdisk1
IBM_2145:ITSO1:superuser>rmvdiskhostmap -host ITSO_AIX itso_aix_hdisk2
IBM_2145:ITSO1:superuser>lshostvdiskmap ITSO_AIX
4. Remove the volumes from the SAN Volume Controller using the svctask rmvdisk
command, which makes the MDisks unmanaged as shown in Example 7-23.
Cached data: When you run the svctask rmvdisk command, the SAN Volume
Controller first double-checks that there is no outstanding dirty cached data for the
volume being removed. If uncommitted cached data still exists, the command fails with
the following error message:
CMMVC6212E The command failed because data in the cache has not been
committed to disk
You must wait for this cached data to be committed to the underlying storage
subsystem before you can remove the volume.
The SAN Volume Controller will automatically destage uncommitted cached data two
minutes after the last write activity for the volume. How much data there is to destage,
and how busy the I/O subsystem is, determine how long this command takes to
complete.
You can check whether the volume has uncommitted data in the cache by using the
svcinfo lsvdisk <VDISKNAME> command and checking the fast_write_state attribute.
This attribute has the following meanings:
empty No modified data exists in the cache.
not_empty Modified data might exist in the cache.
corrupt Modified data might have existed in the cache, but any modified data
has been lost.
Example 7-23 Removing volumes from the SAN Volume Controller

IBM_2145:ITSO1:superuser>rmvdisk itso_aix_hdisk1
IBM_2145:ITSO1:superuser>rmvdisk itso_aix_hdisk2
ctrl_LUN_# controller_name UID tier
4084400500000000 controller0
6005076303ffce63000000000000840500000000000000000000000000000000 generic_hdd
4085400400000000 controller0
6005076303ffce63000000000000850400000000000000000000000000000000 generic_hdd

4085400500000000 controller0
6005076303ffce63000000000000850500000000000000000000000000000000 generic_hdd
8 AIX_MIG01 online unmanaged 16.0GB
0000000000000005 controller1
00173800102f37c4000000000000000000000000000000000000000000000000 generic_hdd
9 AIX_MIG02 online unmanaged 16.0GB
0000000000000006 controller1
00173800102f37c5000000000000000000000000000000000000000000000000 generic_hdd
5. Unmap LUNs from the AIX, and zone the disks from the SAN Volume Controller back to
the AIX server.
Important: This step is the last step that you can perform and still safely back out of
everything you have done so far. Up to this point, you can reverse all of the actions that
you have performed so far to get the server back online without data loss. However,
after you start the next step, you might not be able to turn back without data loss.
To access the LUNs from the AIX server, perform the following steps:
1. Run the cfgmgr -S command to discover the storage subsystem.
2. Use the lsdev -Cc disk command to verify the discovery of the new disk.
3. Remove the references to all of the old disks.
4. If your application and data are on an LVM volume, rediscover the VG and then run the
varyonvg VOLUME_GROUP command to activate the VG.
5. Mount your file systems with the mount /MOUNT_POINT command.
To make sure that the MDisks are removed from the SAN Volume Controller, run the svctask
detectmdisk command. The MDisks are first discovered as offline. Then they will
automatically be removed after the SAN Volume Controller determines that there are no
volumes associated with them.
7.6 SAN Volume Controller Volume migration between two

storage pools
This section addresses the specific steps to perform volume migration for AIX using SAN
Volume Controller between storage pools. If you already have SAN Volume Controller
implemented in environment, using SAN Volume Controller Volume migration functionality is
a non-disruptive way to migrate data between storage pools. For more information about this
process, see 7.3.3, “Migrating a volume between storage pools” on page 192.
If you are implementing SAN Volume Controller for the first time, see 7.2, “SAN Volume
Controller concepts for migrating the data” on page 185.
Using this feature, you can easily migrate the data between two physically different storage
subsystems, fully not apparent to the application.

7.6.1 Environment description
The environment used for this migration example is shown in Figure 7-13. In this
environment, SAN Volume Controller is already in place with back-end storage. New
back-end storage will be added that will be used to migrate the data.
In the example, the AIX system is named ITSO_AIX and the SAN Volume Controller cluster is
named ITSO1. There are two storage zones in use during the migration.
When using the SAN Volume Controller Volume migration feature for data migration, perform
the following basic steps:
1. Create the storage pool from LUNs created on target storage system.
2. Migrate the volumes between the storage pool that holds them to the target storage pool.
3. Check that data is consistent after the migration is finished.
4. Delete the source storage pool if not use in a future. This step is optional.
Important: Volume mirroring method can be used to minimize the impact to the volume
because the volume goes offline only if the source storage pool fails.
Figure 7-13 shows the environment used during testing. As described previously, from the
host side this process is a not apparent and non-disruptive way to migrate data.
Figure 7-13 SAN Volume Controller Volume migration environment
The example uses the following components and tools:

1. AIX 6.1 Server as application host
2. Latest IBM SDD drivers installed

3. IBM SAN Volume Controller, with code release 6.2.0.0
4. Two backend disk subsystems
5. The dd command to simulate read/write I/O operations to file system during migration
6. The nmon tool to collect disk I/O activity for performance data
Tip: Keep your OS level, HBA adapter, SAN environment, SAN Volume Controller code
level, and back-end storage systems up to date to minimize any potential problems.
Table 7-4 shows the AIX host configuration for this test.
Table 7-4 Host specifications

System Model IBM,9119-590
Processor type POWER5
Number of processors 1
Processor clock speed 1654 MHz
CPU type 64 bit
Memory size 1024
Before starting the migration, verify the server information as shown in Example 7-24. The
example uses a single file system built in a volume group. It consists of two LUNs (hdisk41
and hdisk42).
Example 7-24 AIX environment before migration

# lsdev -Cc disk
hdisk41 Available 02-08-02 MPIO FC 2145
# datapath query device

DEV#: 41 DEVICE NAME: hdisk41 TYPE: 2145 ALGORITHM: Load Balance
SERIAL: 60050768018106209800000000000007
===========================================================================
Path# Adapter/Path Name State Mode Select Errors
0* fscsi0/path0 OPEN NORMAL 56 0
2 fscsi0/path2 OPEN NORMAL 30994 0

SERIAL: 60050768018106209800000000000008
===========================================================================

# lspv
hdisk41 00ce0c7b4b252b09 svc_itso_vg active
hdisk42 00ce0c7b4b252c20 svc_itso_vg active
# df -k /svc_itso
file system 1024-blocks Free %Used Iused %Iused Mounted on
/dev/fslv00 26116096 5631780 79% 5 1% /svc_itso
Verify the SAN Volume Controller information as shown in Example 7-25.
Example 7-25 SAN Volume Controller environment

IBM_2145:ITSO1:superuser>lscluster
id name location partnership bandwidth id_alias
0000020060418826 ITSO1 local 0000020060418826
id name status mode mdisk_grp_id mdisk_grp_name capacity ctrl_LUN_#
controller_name UID
tier
0 mdisk1 online managed 1 ITSO_old 16.0GB
0000000000000001 controller1
00173800102f38d6000000000000000000000000000000000000000000000000 generic_hdd
0000000000000002 controller1
00173800102f38d7000000000000000000000000000000000000000000000000 generic_hdd
0000000000000003 controller1
00173800102f38d8000000000000000000000000000000000000000000000000 generic_hdd
0000000000000004 controller1
00173800102f38d9000000000000000000000000000000000000000000000000 generic_hdd
4 mdisk4 degraded_ports unmanaged 15.0GB
4084400400000000 controller0
6005076303ffce63000000000000840400000000000000000000000000000000 generic_hdd
5 mdisk5 online managed 0 ITSO_new 15.0GB
4084400500000000 controller0
6005076303ffce63000000000000840500000000000000000000000000000000 generic_hdd
4085400400000000 controller0
6005076303ffce63000000000000850400000000000000000000000000000000 generic_hdd
4085400500000000 controller0
6005076303ffce63000000000000850500000000000000000000000000000000 generic_hdd

easy_tier_status
0 ITSO_new online 3 2 43.50GB 256 18.50GB
25.00GB 25.00GB 25.00GB 57 80 auto
inactive
1 ITSO_old online 4 0 64.00GB 256 64.00GB
0.00MB 0.00MB 0.00MB 0 80 auto
inactive
id name IO_group_id IO_group_name status mdisk_grp_id mdisk_grp_name
capacity type FC_id FC_name RC_id RC_name vdisk_UID
fc_map_count copy_count fast_write_state se_copy_count
0 AIX_hdisk41 0 io_grp0 online 0 ITSO_new
10.00GB striped 60050768018106209800000000000007 0
1 empty 0
1 AIX_hdisk42 0 io_grp0 online 0 ITSO_new
15.00GB striped 60050768018106209800000000000008 0
1 not_empty 0
IBM_2145:ITSO1:superuser>lshost
id name port_count iogrp_count
0 ITSO_AIX 2 4
1 Blade_sle_5v4 2 4
IBM_2145:ITSO1:superuser>lsvdiskhostmap 0
id name SCSI_id host_id host_name vdisk_UID
0 AIX_hdisk41 0 0 ITSO_AIX 60050768018106209800000000000007
id number_extents
5 14
6 13
7 13
id number_extents
5 20
6 20
7 20
On the SAN Volume Controller side, the storage pools and volumes are already created. For
more information about how to create storage pools and volumes, and how to map volumes to
the hosts, see 7.5, “Migrating SAN disks to SAN Volume Controller volumes and back to SAN”
on page 203 or see Implementing the IBM System Storage SAN Volume Controller V6.1,
SG24-7933.

7.6.2 Performance measurement
To show the performance behavior from the host side, use the dd command to read/write I/O
from the /svc_itso file system. Also, run noon in parallel to collect performance data to show
the performance behavior and any performance impact during the migration phase.
7.6.3 Migration steps

To migrate the data, perform the following steps:
1. Start I/O load to the /svc_itso file system. Example 7-26 shows I/O load from the host
side.
Example 7-26 AIX I/O load

Disk-Adapter-I/O
Name %busy read write xfers Disks Adapter-Type
fcs0 63.1 5287.6 5501.8 KB/s 2697.3 24 FC Adapter
fcs1 73.1 5125.4 5205.5 KB/s 2582.2 24 FC Adapter
vscsi0 1.0 0.0 10.0 KB/s 2.5 1 Virtual SCSI Client A
TOTALS 3 adapters 10413.0 10717.3 KB/s 5282.1 49 TOTAL(MBps)=20.6
Disk-KBytes/second-(K=1024,M=1024*1024)
Disk Busy Read Write Transfers Size Peak% Peak KB/s qDepth
Name KB/s KB/s /sec KB Read+Write or N/A
hdisk41 71% 4796.1 4902.1 2424.6 4.0 75% 12330.5 0
hdisk42 73% 5160.1 5380.1 2635.1 4.0 83% 10949.9 0
Totals(MBps) Read=9.7 Write=10.0 Size(GB)=3983 Free(GB)=18
Figure 7-14 shows I/O load from SAN Volume Controller side.
Figure 7-14 SAN Volume Controller performance view
2. Start the migration process for volumes AIX_hdisk41 and AIX_hdisk42 using the
migratevdisk command as shown in Example 7-27. This process moves the volumes
from the ITSO_old storage pool to the ITSO_new storage pool.
Example 7-27 SAN Volume Controller migratevdisk command

IBM_2145:ITSO1:superuser>svctask migratevdisk -copy 0 -mdiskgrp ITSO_new -vdisk
AIX_hdisk41

IBM_2145:ITSO1:superuser>svctask migratevdisk -copy 0 -mdiskgrp ITSO_new -vdisk
AIX_hdisk42
3. During the migration process, track the progress using the lsmigrate command. You can
also track the extent level migration with the lsvdiskextent command as shown in
Example 7-28.
Example 7-28 SAN Volume Controller migration progress

progress 10
max_thread_count 4
progress 30
max_thread_count 4
id number_extents
0 5
1 5
2 6
3 5
5 7
6 6
7 6
id number_extents
0 12
1 11
2 10
3 11
5 6
6 5
7 5
Tip: The migration process can take some time. The amount depends on the size of your
volumes and the performance of your source and target storage systems. Calculate the
time needed before you start the migration so you have good estimations for your
environment.
Your IBM technical representative can do a performance assessment before you go live
with your migration.

4. When the migration is finished, check the migration status with the lsmigrate and
lsvdiskextent commands as shown in Example 7-29. In the example, lsmigrate shows
that there is no running migration process. The lsvdiskextent command lists the MDisk
extents for the volumes. There are the same number of extents as there were before
starting the migration process.
Example 7-29 SAN Volume Controller migration status

id number_extents
5 14
6 13
7 13
id number_extents
5 20
6 20
7 20
Volumes AIX_hdisk41 and AIX_hdisk42 are moved to new storage pool ITSO_new as
Example 7-30 SAN Volume Controller volumes status

id name IO_group_id IO_group_name status mdisk_grp_id mdisk_grp_name capacity
type FC_id FC_name RC_id RC_name vdisk_UID fc_map_count copy_count
fast_write_state se_copy_count
0 AIX_hdisk41 0 io_grp0 online 0 ITSO_new 10.00GB striped
60050768018106209800000000000007 0 1 empty 0
1 AIX_hdisk42 0 io_grp0 online 0 ITSO_new 15.00GB striped

60050768018106209800000000000008 0 1 not_empty 0
5. Delete storage pool ITSO_old and all assigned Mdisks (Example 7-31).
Example 7-31 SAN Volume Controller removing storage pool

IBM_2145:ITSO1:superuser>rmmdiskgrp -force ITSO_old
Attention: Always be careful when using the -force flag because it forces action
without any warning.

6. From the AIX side, check the volume, physical volumes, and file systems. Everything
should remain the same after migration is complete and the storage pool ITSO_old is
deleted as shown in Example 7-32.
Example 7-32 AIX after migration is finished

# lsdev -Cc disk
# datapath query device

SERIAL: 60050768018106209800000000000007
===========================================================================

SERIAL: 60050768018106209800000000000008
===========================================================================
# lspv
hdisk41 00ce0c7b4b252b09 svc_itso_vg active
hdisk42 00ce0c7b4b252c20 svc_itso_vg active
# df -k /svc_itso
file system 1024-blocks Free %Used Iused %Iused Mounted on
/dev/fslv00 26116096 26110500 1% 6 1% /svc_itso
7.6.4 Performance Analyses

This section highlights the performance impact during migration in the example. The graphs
are divided into the following sections:
򐂰 Old Storage pool (ITSO_old) before migration
򐂰 Migration
򐂰 New Storage pool (ITSO_new) after migration

You can also see where the I/O load was active and when it was stopped.
Figure 7-15 shows overall System Summary for ITSO_AIX during the example migration.
Figure 7-15 SAN Volume Controller performance overview: ITSO_AIX System summary
Figure 7-16 shows the Disk Busy graph for ITSO_AIX.
Figure 7-16 SAN Volume Controller performance overview: ITSO_AIX Host Disk Busy
Migration using SAN Volume Controller does affect the host I/O performance in the example
environment, but only for a short time. This performance impact is closely connected to the
back-end disk subsystem and its performance profile.
The test used the default number of threads (four) on SAN Volume Controller. Four threads is
the maximum value and it can saturate back-end disk system, as you can see in the example.
However, the process can be prioritized by specifying the number of threads to be used in
parallel (from 1 to 4) while migrating. Using a single thread creates the least background load
on the system.
You can also see in the graphs that the data was migrated from a higher performance to a
lower performance tier storage.

7.7 Data migration using SAN Volume Controller mirrored
volumes
This section shows how to use the volume mirroring feature in SAN Volume Controller to
migrate data from one storage device to another. If you already have SAN Volume Controller
implemented in your environment, using the SAN Volume Controller mirroring volumes
function is a non-disruptive way to migrate data.
Using this feature, you can easily migrate data between two physically separate storage
subsystems. The process is fully not apparent to the application. Volume mirroring is included
in the base virtualization license, so you do not need to purchase any additional functions for
the SAN Volume Controller. This function can also be used to migrate from a fully allocated
volume to a thin-provisioned volume.
When using volume mirroring, the zero detect feature for thin-provisioned volumes allows you
to reclaim unused allocated disk space (zeros) when converting a fully allocated volume to a
thin-provisioned volume.
Using SAN Volume Controller mirrored volumes feature for data migration involves the
following basic steps:
1. Add the target copy for source volume.
2. Run synchronization and wait for it to complete.
3. Remove the source volume.
Important: When using SAN Volume Controller mirroring volume features, the data
migration can be stopped at any time without compromising data consistency on the
primary volumes.
7.7.1 Environment description

Figure 7-17 on page 238 shows the environment used during testing. Operating system
environments are not covered because if SAN Volume Controller is already in place, the
procedure is the same for any operating system.
The example uses the following components and tools:

1. Windows Server 2008 as application host
2. IBM SAN Volume Controller with code release 6.2.0.0
3. Two backend disk subsystems (source and target)
4. Iometer to simulate read/write operations to logical volume during mirroring
5. Windows Performance monitor to track disk performance and availability from the host
side

Figure 7-17 SAN Volume Controller mirrored volume lab setup
The SAN Volume Controller side has a storage pool created from an old disk subsystem
called ITSO_old. The target storage pool is on a new disk subsystem and called ITSO_new.
From a migration point of view, the process is similar to the one described in 7.6, “SAN
Volume Controller Volume migration between two storage pools” on page 227. However, the
two volumes are kept in synchronization after migration is complete.
The SVCs mirrored volume background architecture has the following characteristics:
򐂰 By default, reads will still be serviced from the original volume, but this setting can be
changed after the two volumes are synchronized.
򐂰 All writes are sent to both volumes synchronously.
򐂰 It is possible to split the two volumes at a defined point in time. This characteristic allows
you to schedule a controlled split, such as when the host is powered off.
򐂰 Allows a defined point in time copy to be taken.
򐂰 The second copy of the volume can be removed at any time, allowing easy regression.

Figure 7-18 shows the storage pools on SAN Volume Controller ITSO1 as shown in the GUI.
Figure 7-18 SAN Volume Controller ITSO storage pools
Example 7-33 shows the storage pools on SAN Volume Controller ITSO1 shown in the
command-line interface.
Example 7-33 SAN Volume Controller ITSO1 storage pools from CLI
IBM_2145:ITSO1:superuser>svcinfo lsmdiskgrp
virtual_capacity used_capacity real_capacity
0 ITSO__old online 3 1 45.00GB 256 35.00GB
10.00GB 10.00GB 10.00GB
1 ITSO__new online 4 1 62.50GB 256 52.50GB
10.00GB 10.00GB 10.00GB
Figure 7-19 shows the SAN Volume Controller ITSO1 volumes from the GUI.
Figure 7-19 SAN Volume Controller volume information
Example 7-34 shows the storage pools on SAN Volume Controller ITSO1 shown in the
command-line interface.
Example 7-34 SAN Volume Controller ITSO1 volumes from CLI

IBM_2145:ITSO1:superuser>svcinfo lsvdisk
id name IO_group_id IO_group_name status capacity vdisk_UID
0 sle_5v4_vol01 0 io_grp0 online 10.00GB
60050768018106209800000000000003

Example 7-35 shows host system with the volume mapped from old storage using SAN
Volume Controller.
Example 7-35 SAN Volume Controller volume mapped to host.

C:\Program Files\IBM\SDDDSM>datapath.exe query device
Total Devices : 1
DEV#: 0 DEVICE NAME: Disk1 Part0 TYPE: 2145 POLICY: OPTIMIZED
SERIAL: 60050768018106209800000000000003
============================================================================
Path# Adapter/Hard Disk State Mode Select Errors
0 Scsi Port1 Bus0/Disk1 Part0 OPEN NORMAL 0 0
To simulate I/O load on the host side, Iometer created constant I/O load on the volume used
for migration. Figure 7-20 shows the Iometer window with the simulated I/O load.
Figure 7-20 Iometer window

Figure 7-21 shows that the I/O load exists on the SAN Volume Controller side.
Figure 7-21 SAN Volume Controller performance view
Tip: The steps described in this section are the same regardless of the host operating
system.
7.7.2 Creating mirrored volumes using the SAN Volume Controller GUI
To create mirrored volumes using the GUI, perform the following steps:
1. Click Volumes All Volumes.
2. Right-click the volume and select Volume Copy Actions Add Mirrored Copy
(Figure 7-22).
Figure 7-22 SAN Volume Controller adding mirrored copy

3. Select ITSO_new as shown on Figure 7-23 and click Add Copy.
Figure 7-23 SAN Volume Controller Mirrored Copy confirmation window
Tip: You can select Thin Provisioning option on this window.
Figure 7-24 shows volume copies after adding mirrored copy.
Figure 7-24 SAN Volume Controller volume with mirrored copy
Note: The asterisk (*) shows that Copy1 is the primary volume.

After you add the mirrored copy, the synchronization process starts immediately
(Figure 7-25). When the synchronization finishes, you can change the primary copy
volume from Copy1 to Copy0.
Figure 7-25 SAN Volume Controller Synchronization progress
4. When synchronization is complete, click Volumes All Volumes.

5. Right-click Copy0 and select Make Primary as shown on Figure 7-26.
Figure 7-26 SAN Volume Controller Make Primary copy
Volume sle_5v4_vol01 changes storage pool location from ITSO_old to ITSO_new, but
retains the same UID number as shown in Figure 7-27.
Figure 7-27 SAN Volume Controller Mirrored Copy window after we change Primary Copy
6. Click Volumes All Volumes.

7. Right-click Copy1 from the ITSO_old storage pool and select Delete this Copy as shown
on Figure 7-28.
Figure 7-28 SAN Volume Controller deleting mirrored volumes
8. Click OK to confirm the deletion.

The migration process is complete and you are using the new disk subsystem. Figure 7-29
shows the status of volume sle_5v4_vol01, which is placed on the new storage pool called
ITSO_new.
Figure 7-29 SAN Volume Controller volume after migration
Tip: SAN Volume Controller supports migration from virtualized MDisks and from image
mode MDisks.
7.7.3 Creating mirrored volumes using CLI

This section addresses all procedures to migrate data using CLI commands. To do so,
1. Add a mirrored volume copy for existing volume sle_5v4_vol01 using the addvdiskcopy
command (Example 7-36). Adding a copy to an existing volume changes a non-mirrored
volume into a mirrored volume.
Example 7-36 SAN Volume Controller adding mirrored volume copy

IBM_2145:ITSO1:superuser>svcinfo lsvdisk
id name IO_group_id IO_group_name status mdisk_grp_id mdisk_grp_name capacity
type FC_id FC_name RC_id RC_name vdisk_UID fc_map_count copy_count
fast_write_state se_copy_count
0 sle_5v4_vol01 0 io_grp0 online 0 ITSO__old 10.00GB striped
60050768018106209800000000000003 0 1 not_empty 0
easy_tier_status
0 ITSO__old online 3 1 45.00GB 256 35.00GB 10.00GB 10.00GB 10.00GB 22 80 auto
inactive

1 ITSO__new online 4 0 62.50GB 256 62.50GB 0.00MB 0.00MB 0.00MB 0 80 auto
inactive
IBM_2145:ITSO1:superuser>svctask addvdiskcopy -mdiskgrp 1 0

Vdisk [0] copy [0] successfully created
IBM_2145:ITSO1:superuser>svcinfo lsvdiskcopy
vdisk_id vdisk_name copy_id status sync primary mdisk_grp_id mdisk_grp_name
capacity type se_copy easy_tier easy_tier_status
0 sle_5v4_vol01 0 online no no 1 ITSO__new 10.00GB striped no on inactive
0 sle_5v4_vol01 1 online yes yes 0 ITSO__old 10.00GB striped no on inactive
IBM_2145:ITSO1:superuser>lsvdisksyncprogress
vdisk_id vdisk_name copy_id progress estimated_completion_time
0 sle_5v4_vol01 0 11 110719235627
IBM_2145:ITSO1:superuser>lsvdisksyncprogress
2. When the synchronization process is finished, change the primary volume from the
ITSO_old storage pool to ITSO_new as shown in Example 7-37.
Example 7-37 SAN Volume Controller change primary copy

IBM_2145:ITSO1:superuser>lsvdiskcopy
0 sle_5v4_vol01 0 online yes yes 1 ITSO__new 10.00GB striped no on inactive
0 sle_5v4_vol01 1 online yes no 0 ITSO__old 10.00GB striped no on inactive
IBM_2145:ITSO1:superuser>svctask chvdisk -primary 0 0

3. Delete the copy from the storage pool ITSO_old as shown in Example 7-38.
Example 7-38 SAN Volume Controller change primary copy

IBM_2145:ITSO1:superuser>lsvdiskcopy
0 sle_5v4_vol01 0 online yes yes 1 ITSO__new 10.00GB striped no on inactive
0 sle_5v4_vol01 1 online yes no 0 ITSO__old 10.00GB striped no on inactive
IBM_2145:ITSO1:superuser>svctask rmvdiskcopy -copy 1 0

Further information: For more information about CLI commands, see the Command-Line
Interface User's Guide, Version 6.2.0, GC27-2287-01.

7.7.4 Performance analysis
Figure 7-30 shows the impact during the migration process. The performance is normal
during migration phase except two short performance peaks.One is when you add the
mirrored copy, and the second while changing the primary volume.
This example is from a testing environment. In a production environment, pre-migration

planning is crucial for successful migration.
Figure 7-30 Performance view
7.8 Data migration using SAN Volume Controller Metro Mirror

The SAN Volume Controller Metro Mirror can be used to migrate data while the application
server is still active on the source storage (through the source SAN Volume Controller). The
source SAN Volume Controller cluster in a Metro Mirror partnership is called the Master SAN
Volume Controller cluster. The target SAN Volume Controller cluster is called the Auxiliary
SAN Volume Controller cluster. This section outlines the steps to set up a Metro Mirror
partnership, relationships and, optionally, Metro Mirror consistency groups
In this example, a Metro Mirror relationship between two SAN Volume Controller clusters is
used to demonstrate how the copy services function can migrate data between two physically
separate locations. You can also use Metro Mirror as a data migration solution within a single
SAN Volume Controller cluster.

Figure 7-31 shows the test environment.
Figure 7-31 SAN Volume Controller Metro Mirror Lab Setup
7.8.1 SAN Volume Controller Metro Mirror partnership

The first step is to create a Metro Mirror partnership between the clusters that you want to
migrate between. This section contains the steps for the following procedures:
򐂰 Creating a Metro Mirror partnership using the SAN Volume Controller GUI
򐂰 Creating a Metro Mirror partnership using the SAN Volume Controller CLI
Creating a Metro Mirror partnership using the SAN Volume Controller

GUI
Create an SAN Volume Controller Metro Mirror partnership between the two clusters using
1. Click Copy Services  Partnerships New Partnership.
2. On the Create Partnership window, define the remote cluster as shown in Figure 7-32 on
page 248. In this example, there is a single remote cluster named ITSO2. Select the
cluster, specify the wanted bandwidth in MBps, and then click Create.

Figure 7-32 Choose the remote cluster
Important: Set the bandwidth used by the background copy process to less than or
equal to the bandwidth that can be sustained by the communication link between the
systems. The link must be able to sustain any host requests and the rate of background
copy.
If the -bandwidth parameter is set to a higher value than the link can sustain, the
background copy process uses the actual available bandwidth.
3. The final window shows a listing of the partnership just created as shown in Figure 7-33.
The status of the partnership is Fully Configured, which means that the partnership has
been created from both clusters. If this cluster is the first cluster to define the partnership,
the status is listed as Partially Configured. Repeat this process from the other cluster to
get to the Fully Configured state.
Figure 7-33 Mirror cluster partnership

Creating a Metro Mirror partnership using the SAN Volume Controller
CLI
You can also use SAN Volume Controller CLI commands to create the Metro Mirror
partnership using the following steps:
1. Verify that the source and target SAN Volume Controller clusters can communicate by
issuing the svcinfo lsclustercandidate command as shown in Example 7-39. In this
example, the command is issued from the cluster named ITSO1. The candidate cluster
name is ITSO2.
Example 7-39 List the candidate clusters f

IBM_2145:ITSO1:admin>svcinfo lsclustercandidate
id configured cluster_name
00000200604187DE no ITSO2
IBM_2145:ITSO2:superuser>svcinfo lsclustercandidate
id configured name
0000020060418826 yes ITSO1
2. To create the partnership between the two clusters, run the svctask mkpartnership
command from each cluster (Example 7-40). The bandwidth parameter is mandatory.
Example 7-40 Create partnership

IBM_2145:ITSO1:superuser>svctask mkpartnership -bandwidth 50 00000200604187DE
IBM_2145:ITSO2:superuser>svctask mkpartnership -bandwidth 50 0000020060418826
3. Verify the creation of the partnership using the svcinfo lscluster command as shown in
Example 7-41. The partnership is displayed as partially configured if only one SAN
Volume Controller cluster has created the partnership.
Example 7-41 Verify the SAN Volume Controller partnership

IBM_2145:ITSO1:superuser>svcinfo lscluster
0000020060418826 ITSO1 local 0000020060418826
00000200604187DE ITSO2 remote fully_configured 50 00000200604187DE
IBM_2145:ITSO2:superuser>svcinfo lscluster
00000200604187DE ITSO2 local 00000200604187DE
0000020060418826 ITSO1 remote fully_configured 50 0000020060418826
7.8.2 SAN Volume Controller Metro Mirror relationships

After you establish the Metro Mirror partnership, you are ready to create the mirror
relationships. This section contains the steps for the following procedures:
򐂰 Creating Metro Mirror relationships using SAN Volume Controller GUI
򐂰 Creating Metro Mirror relationships using SAN Volume Controller CLI

Creating Metro Mirror relationships using SAN Volume Controller GUI
To create the Metro Mirror relationships using the GUI, perform the following steps:
1. Click Copy Services  Remote Copy New Relationship.
2. On the New Relationship window, select Metro Mirror and click Next (Figure 7-34).
Figure 7-34 Steps for creating Metro Mirror relationships
3. Select the Auxiliary Cluster ITSO02 (on another system) where the volumes are located
and click Next (Figure 7-35).
Figure 7-35 Select the auxiliary cluster

4. Select the master and auxiliary volume pair for the Metro Mirror. The master and auxiliary
volume must be the same size. Only volumes that are suitable for the relationship are
listed. In this example, select ITSO_Auxiliary as shown in Figure 7-36. Click Add and
then Next.
Figure 7-36 Select the Auxiliary Volume
5. Select a synchronization option for the volumes (Figure 7-37). In this example, select No
because this is the status of the relationship. Click Next.
Figure 7-37 Select synchronization status
6. Select Yes to start replication immediately (Figure 7-38), and click Next.
Figure 7-38 Select to start copying process

7. Track the synchronization process from the GUI by clicking Running task from the lower
right window as shown in Figure 7-39.
Figure 7-39 Remote Copy progress
Creating Metro Mirror relationships using SAN Volume Controller CLI

When creating a relationship using the SAN Volume Controller CLI, you can create two
different types of Metro Mirror relationships:
򐂰 Associated with a consistency group: Consistency groups preserve data consistency
across multiple Metro Mirrored volumes. Use them when your applications have related
data that span multiple volumes.
򐂰 Stand-alone relationships: Relationships can be created outside a consistency group.
Creating relationships as part of a consistency group

To create a Metro Mirror relationship within a consistency group, perform the following steps:
1. Create the consistency group using the svctask mkrcconsistgrp command
(Example 7-42). In this example, the group is named Lnxsmallfs_cg and is being issued
on local cluster ITSO1. The remote cluster is ITSO2. The svcinfo lsrcconsistgrp
command verifies creation.
Example 7-42 Create an empty consistency group

IBM_2145:ITSO1:superuser>svctask mkrcconsistgrp -name Lnxsmallfs_cg -cluster
ITSO2
RC Consistency Group, id [0], successfully created
IBM_2145:ITSO1:superuser>svcinfo lsrcconsistgrp
id name master_cluster_id master_cluster_name aux_cluster_id
aux_cluster_name primary state relationship_count copy_type
0 Lnxsmallfs_cg 0000020060418826 ITSO1 00000200604187DE ITSO2
empty 0 empty_group
2. After a consistency group is created, create Metro Mirror relationships as part of that
consistency group. Example 7-43 shows the Metro Mirror relationship named
lnx_smallfs_mm, created as part of consistency group Lnxsmallfs_cg. It is between
volume ITSO_Master on the source cluster and ITSO_Auxiliary on the target cluster. The
command svctask mkrcrelationship creates the Metro Mirror relationship and svcinfo
lsrcrelationship verifies its creation.
Example 7-43 Create a Metro Mirror relationship associated with a consistency group
IBM_2145:ITSO1:superuser>svctask mkrcrelationship -master ITSO_Master -aux
ITSO_Auxiliary -name lnx_smallfs_mm -consistgrp Lnxsmallfs_cg -cluster ITSO2

RC Relationship, id [1], successfully created
IBM_2145:ITSO1:superuser>svcinfo lsrcrelationship 1
id 1
name lnx_smallfs_mm
master_cluster_id 0000020060418826
master_cluster_name ITSO1
master_vdisk_id 1
master_vdisk_name ITSO_Master
aux_cluster_id 00000200604187DE
aux_cluster_name ITSO2
aux_vdisk_id 0
aux_vdisk_name ITSO_Auxiliary
primary master
consistency_group_id 0
consistency_group_name Lnxsmallfs_cg
state inconsistent_stopped
bg_copy_priority 50
progress 0
freeze_time
status online
sync
copy_type metro
Creating stand-alone relationships

If the Metro Mirror relationships do not need to be part of a consistency group, stand-alone
relationships can be created using the same svctask rmkrcrelationship command. The
-consistgrp parameter can be omitted in this case.
Example 7-44 shows the creation of a stand-alone relationship named lnx_mediumfs_mm

between volumes ITSO_Master and ITSO_Auxiliary.
Example 7-44 Create a stand-alone Metro Mirror relationship

IBM_2145:ITSO1:superuser>svctask mkrcrelationship -master ITSO_Master -aux
ITSO_Auxiliary -name lnx_mediumfs_mm -cluster ITSO2
RC Relationship, id [1], successfully created
IBM_2145:ITSO1:superuser>svcinfo lsrcrelationship 1
id 1
name lnx_mediumfs_mm
master_vdisk_id 1
aux_vdisk_id 0
primary master
consistency_group_id
consistency_group_name
state inconsistent_stopped
bg_copy_priority 50
progress 0

freeze_time
status online
sync
copy_type metro
7.8.3 Starting and monitoring SAN Volume Controller Metro Mirror Copy
The next step in the data migration is to copy the data to the target disk. This section contains
the steps for the following procedures:
򐂰 Starting Metro Mirror Copy using SAN Volume Controller GUI
򐂰 Start Metro Mirror Copy using SAN Volume Controller CLI
Starting Metro Mirror Copy using SAN Volume Controller GUI

To copy the data using the SAN Volume Controller GUI, perform the following steps:
1. Click Copy Services  Remote Copy.
2. Select the Metro Mirror relationship (Figure 7-40). The state Inconsistent Stopped means
that this relationship has not been started. In this example, the data on both Auxiliary
volumes are not yet synchronized. Click Actions  Start.
Figure 7-40 Start Copy process
After the data is copied, the state for the two Volumes is Consistent Synchronized as
Figure 7-41 Consistent Synchronized state
Start Metro Mirror Copy using SAN Volume Controller CLI

Metro Mirror consistency groups and stand-alone relationships are started and monitored
with two sets of commands.

Copying data in consistency groups
Use the svctask startrcconsistgrp command to start the Metro Mirror copy for all the
relationships in the consistency group (Example 7-45). The command svcinfo
lsrcconsistgrp shows a state of inconsistent_copying. This state means that the copy is in
progress. The command svcinfo lsrcrelationships shows the progress of the copy.
Example 7-45 Start and monitor a Metro Mirror consistency group

IBM_2145:ITSO1:admin>svctask startrcconsistgrp Lnxsmallfs_cg
IBM_2145:ITSO1:superuser>svcinfo lsrcconsistgrp Lnxsmallfs_cg

id 0
name Lnxsmallfs_cg
primary master
state inconsistent_copying
relationship_count 1
freeze_time
status
sync
copy_type metro
RC_rel_id 1
RC_rel_name lnx_smallfs_mm
IBM_2145:ITSO1:superuser>svcinfo lsrcrelationship lnx_smallfs_mm

id 1
name lnx_smallfs_mm
master_vdisk_id 1
aux_vdisk_id 0
primary master
bg_copy_priority 50
progress 9
freeze_time
status online
sync
copy_type metro
After all the pairs are synced, the consistency group state is consistent_synchronized
(Example 7-46).
Example 7-46 Consistency group that has been synchronized

IBM_2145:ITSO1:admin>svcinfo lsrcconsistgrp Lnxsmallfs_cg
id 0

name Lnxsmallfs_cg
primary master
state consistent_synchronized
freeze_time
status
sync
copy_type metro
RC_rel_id 1
IBM_2145:ITSO1:superuser>svcinfo lsrcrelationship lnx_smallfs_mm

id 1
name lnx_smallfs_mm
master_vdisk_id 1
aux_vdisk_id 0
primary master
bg_copy_priority 50
progress
freeze_time
status online
sync
copy_type metro
Copying data in stand-alone relationships

Stand-alone relationships are started using the svctask startrcrelationship command.
Again, the progress of the copy can be monitored using the svcinfo lsrcrelationship
command.
Example 7-47 Metro Mirror stand-alone relationship

IBM_2145:ITSO1:admin>svctask startrcrelationship lnx_mediumfs_mm
IBM_2145:ITSO1:superuser>svcinfo lsrcrelationship lnx_mediumfs_mm

id 1
master_vdisk_id 1

aux_vdisk_id 0
primary master
bg_copy_priority 50
progress 3
freeze_time
status online
sync
copy_type metro
IBM_2145:ITSO1:admin>svcinfo lsrcrelationship lnx_mediumfs_mm

id 1
master_vdisk_id 1
aux_vdisk_id 0
primary master
bg_copy_priority 50
progress
freeze_time
status online
sync
copy_type metro
7.8.4 Stopping SAN Volume Controller Metro Mirror Copy

After the mirror relationships reach a synchronized state, move the application to the new
storage subsystem. There are two options available:
򐂰 Keep SAN Volume Controller in the configuration and connect the application to virtualized
storage.
򐂰 Remove SAN Volume Controller and attach the application to the new storage subsystem.
After you migrate from the old storage to the new storage using SVC Metro Mirror, you must
switch your application to the new storage. This requires the application to be shut down
because you have to switch application access from old to new storage.
Before starting these steps, shut down the application and preparing the system to move to
the new storage. In general, these steps include:
򐂰 Quiesce all I/O.
򐂰 Shut down the application server.
򐂰 Stop the Metro Mirror copy.

򐂰 Remove the Metro Mirror relationships.
򐂰 Start the application server on the new storage, with or without SAN Volume Controller.
Stopping Metro Mirror using SAN Volume Controller GUI

To stop Metro Mirror copy using the SAN Volume Controller GUI, perform the following steps:
1. From the Master cluster, click Copy Services  Remote Copy.
2. Select the consistency group (Lnxsmallfs_cg in this example) and click Action  Stop to
stop Metro Mirror copy process as shown in Figure 7-42.
Figure 7-42 Stop Copy Process
3. Enable write access to the target disk. Be sure to select Allow secondary read/write
access as shown in Figure 7-43. Selecting this option allows the application server to
access the target volumes as the new source volumes. Click Stop Consistency Group to
stop the copy.
Figure 7-43 Enable write access

The state of the relationship is now displayed as Idling. The relationship still exists, but no
data is being copied. Figure 7-44 shows the status of the relationships in the example, In
Sync/Idling, after the copy is stopped.
Figure 7-44 Viewing relationships in the idle state
4. Select the consistency group (Lnxsmallfs_cg in the example) and click Delete as shown in
Figure 7-45.
Figure 7-45 Delete Metro Mirror relationships

6. Delete the partnership between Master volume and Auxiliary volume by clicking Copy
Services  Remote Copy  Delete Partnership as shown in Figure 7-46.
Figure 7-46 Confirm delete
Stopping Metro Mirror using SAN Volume Controller CLI

To stop Metro Mirror copy using the SAN Volume Controller CLI, perform the following steps:
1. Stop the consistency groups using the svctask stoprcconsistgrp command with the
-access parameter as shown in Example 7-48.
Example 7-48 Stop a consistency group and allow access to the target VDisk
IBM_2145:ITSO1:admin>svctask stoprcconsistgrp -access Lnxsmallfs_cg
IBM_2145:ITSO1:superuser>svcinfo lsrcconsistgrp Lnxsmallfs_cg

id 0
name Lnxsmallfs_cg
primary
state idling
freeze_time
status
sync in_sync
copy_type metro
RC_rel_id 1

2. Stop the stand-alone Metro Mirror relationships and allow host access to the target
volumes using the svctask stoprcrelationship command with the -access parameter
(Example 7-49).
Example 7-49 Stop Metro Mirror relationships

IBM_2145:ITSO1:superuser>svctask stoprcrelationship -access lnx_smallfs_mm
IBM_2145:ITSO1:admin>svcinfo lsrcrelationship lnx_smallfs_mm
id 1
name lnx_smallfs_mm
master_vdisk_id 1
aux_vdisk_id 0
primary
state idling
bg_copy_priority 50
progress
freeze_time
status
sync in_sync
copy_type metro
7.8.5 Performance overview

This section addresses the performance impact during Metro Mirror synchronization for the
example environment. A heavy I/O load was added during testing to simulate a real case
scenario.
Figure 7-47 shows overall host System Summary during the Metro Mirror process.
Figure 7-47 SAN Volume Controller host performance system summary

Figure 7-48 shows host Disk Busy graph.
Figure 7-48 SAN Volume Controller host performance Disk utilization
The testing scenarios showed that Metro Mirror replication using SAN Volume Controller had
no impact on the host I/O performance. This allowed sustained I/O load during Metro Mirror
replication.
However, when planning for Metro Mirror replication, be sure that you use correct link sizing.
Insufficient link bandwidth can saturate source disks. Contact your IBM technical
representative to help you properly size the replication link.
7.9 SAN Volume Controller as data migration engine

The primary use of the SAN Volume Controller is not as a storage migration tool. However,
the advanced capabilities of the SAN Volume Controller allow you to use the SAN Volume
Controller to migrate data. Therefore, you can add the SAN Volume Controller temporarily to
your SAN environment to copy the data from one storage subsystem to another storage
subsystem.
The SAN Volume Controller allows you to copy image mode volumes directly from one
subsystem to another subsystem while host I/O is running. The only downtime required is
when the SAN Volume Controller is added to and removed from your SAN environment. This
scenario is described in 7.5.6, “Preparing to migrate from the SAN Volume Controller” on
page 220.
To use SAN Volume Controller for migration purposes only, perform the following steps:
1. Add SAN Volume Controller to your SAN environment.
2. Configure the SAN Volume Controller to fits your needs.
3. Prepare your application for data migration (unmount file systems and detach LUNs).
4. Add SAN Volume Controller between your storage and the host.
5. Create image mode disks on SAN Volume Controller from the LUNs that you migrate.
6. Attach LUNs, mount the file systems, and start the application.
7. Start the migration.
8. After the migration process is complete, detach the selected LUNs.

9. Remove SAN Volume Controller from your SAN.
10.Attach the LUNs, and start the application.
As you can see, little downtime is required. If you prepare everything correctly, you should be
able to reduce your downtime to a few minutes. The copy process is handled by the SAN
Volume Controller, so there is no performance impact on the host during the migration
process.
7.10 Other resources

For more information about SAN Volume Controller based migration, see the following
publications:
򐂰 Implementing the IBM System Storage SAN Volume Controller V6.1, SG24-7933
򐂰 Software Installation and Configuration Guide Version 6.2.0, GC27-2286-01
򐂰 Command-Line Interface User's Guide Version 6.2.0, GC27-2287-01
򐂰 Official SAN Volume Controller guides can be found at:
http://publib.boulder.ibm.com/infocenter/svc

8
Chapter 8. Using mirroring techniques

This chapter addresses migrating data from a disk storage source device to a DS8000 target
device using Logical Volume Manager (LVM)-based mirroring (split mirroring). This chapter
contains the following sections:
򐂰 Concepts of the LVM
򐂰 Preparation and planning
򐂰 Migrating data using Windows Logical Disk Manager
򐂰 Data migration using Solaris Volume Manager
򐂰 Data migration using Veritas Storage Foundation
򐂰 Data migration using HP-UX Volume Manager mirroring
򐂰 Data migration using AIX LVM mirroring
򐂰 Data migration in a PowerHA clustered environment using AIX LVM mirroring
򐂰 Data migration using Linux LVM2 mirroring
򐂰 Data migration using network block devices

8.1 Concepts of the LVM
Logical Volume Manager (LVM) is available for most open systems. For AIX and Linux, it is
called LVM. For Solaris and HP-UX, it can be known as LVM or VxVM. For the Windows
platform, it is known as Logical Disk Manager (LDM). The LVM or LDM creates a layer of
virtualization within the operating system.
Every volume management software provides the following basic functions:

򐂰 Extend logical volumes across several physical disks
򐂰 Stripe data across several physical disks to improve performance
򐂰 Mirror data for high availability and migration
The LUNs provided by a DS8000 are presented to the LVM or LDM as physical SCSI disks.
The normal process for a migration is to set up a mirror of the data on the old disks to the new
LUNs. You wait until they are synchronized, and then split them at the cut-over time. Some
volume management software provides commands that automate this process.
Disk mirroring was first mentioned in a 1978 patent awarded to Norman Ken Ouchi of the IBM
Corporation. Logical Volume mirroring came into general use in UNIX operating systems
around the mid 1990s. At that time, it was used primarily for data protection. However, the
“split mirror” technique of moving data from one disk to another became a handy tool for the
system administrator.
A lot has changed in the last 10 years, but LVM Mirroring is still an effective way to get data
from one LUN to another. The process is straightforward. You have a logical volume on a LUN
(data source) with data that you want to relocate or migrate to a LUN on a DS8000 (data
destination). The process involves these steps:
1. Configure the destination LUN
2. Have the operating system recognize the LUN
3. Use the operating system LVM to create a logical volume on the DS8000 LUN
4. Establish a mirror between the source and destination logical volumes
5. Sync the mirror
6. Split the mirror
The data is now on the destination LUN.
The application using the source logical volume or file system does not need to be stopped
during the LV mirror migration process. The process is non-disruptive as long as the operating
system allows you to add and remove devices dynamically. However, typically you want the
application to use the data on the destination logical volume after the migration. In this case,
the application must be quiesced and reconfigured to use the destination logical volume. For
a completely non-disruptive way to migrate data, use solutions such as those provided by
Softek TDMF or zDMF.
A large advantage of using the volume management software for data migration is that it also
allows for various types of consolidation. It can do this due to the virtualization nature of
volume management software,

The major disadvantage of the volume management software mirroring method is that it
requires substantial system administration, intervention, and attention. Production data is
manipulated while production is running, and so migration requires host cycles.
Important: If you are planning to use volume management software functions to migrate
data, be careful with limitations. These limitations include the total number of physical disks
in the same volume group or volume set. In addition, if you are consolidating volumes with
different sizes, check the procedures to see whether consolidation is possible.
8.2 Preparation and planning

As highlighted in Chapter 2, “Migration techniques and processes” on page 7, any migration
project must start with a preparation and planning phase. Create a diagram of the overall
environment (Figure 8-1 on page 269) and create a list of all the components in your
environment.
Check the compatibility of the following components in your environment:

򐂰 Prepare the host servers. Verify that both the host and host HBAs can connect to the
existing storage (source) and the DS8000 (target). See the Interoperability matrix available
at:
http://www.ibm.com/systems/support/storage/config/ssic/index.jsp
– Check vendor and model type (change or add HBAs, if necessary).
– Check the HBA firmware level and upgrade if necessary. Supported HBA firmware
levels are available at the System Storage Interoperation Center (SSIC):
http://www-03.ibm.com/systems/support/storage/ssic/interoperability.wss
– Determine the number and size of each source volume (LUN) on each host.
– Determine how the LUN is spread across multiple disks for performance
considerations.
– Determine whether the existing multipath driver can coexist with the supported host
disk driver such as SDD, or MPIO.
򐂰 Prepare the DS8000 arrays and LUNs to provide the best possible performance.
– Select RAID types, for example RAID-5 or RAID-10
– Consider LUN to array mapping considerations for sharing, spreading, or isolating the
type of requirements
– Make sure that the latest firmware is loaded. You can verify the current level against the
latest level at:
https://ssgtech3.tucson.ibm.com/S96A/DS8000%20PE%20Field%20Tips.nsf/wvByDate
Modified?OpenView
http://www-03.ibm.com/systems/storage/software/virtualization/svc/interop.ht
ml
򐂰 Prepare the SAN fabric.
– Check fabric compatibility for connection to both the existing storage server and the
DS8000.
Chapter 8. Using mirroring techniques 267

• Verify that fabric firmware levels are supported with the DS8000 (upgrade if
necessary). For specific support, see the following Fibre Channel Director websites:
Cisco SAN:
http://www-1.ibm.com/servers/storage/san/c_type/
CNT (Inrange) SAN:
http://www-1.ibm.com/servers/storage/san/n_type/
IBM SAN (Brocade) SAN:
http://www-1.ibm.com/servers/storage/san/b_type/
McData SAN:
http://www-1.ibm.com/servers/storage/san/m_type/
• Verify that you have enough free ports to connect the DS8000 nodes to the fabric.
Add more if needed.
– Verify the cabling.
• Order the correct number of cables.
• Connect the DS8000 to the fabric.
• Provide cabling diagrams to the cabling vendor or the person responsible for the
cabling.
– Verify the zoning.
• Zone the host HBAs to the DS8000.
򐂰 Prepare the existing storage server.
– Verify that the existing storage server is in optimal condition.
• Check that all connections, ports, arrays, and LUNs are functioning properly.
• Check firmware levels (upgrade if necessary).
– Discover and document the logical and physical layout of the arrays and LUN
mappings to the hosts.
8.3 Migrating data using Windows Logical Disk Manager

Dynamic disks were first introduced with the Windows 2000 server and provide features that
basic disks do not. One of these features is the ability to create fault-tolerant volumes. This
section addresses how to create a mirror using the Logical Disk Manager with dynamic disks.
8.3.1 Sample Windows LDM migration test environment

This section uses the test environment shown in Figure 8-1 on page 269. It is composed of
the following components:
򐂰 Windows 2008 Enterprise Edition, IBM BladeCenter LS21
– Disk device drivers:
• Microsoft MPIO Version 6.0.6001.18000
• IBM Subsystem Device Driver DSM Version 2.4.3.1-1
– Emulex based LP1105 HBAs running firmware version 2.70A5
򐂰 IBM XIV running microcode level 10.2

򐂰 DS8300 model 921 running firmware level SEA 5.4.3.65, bundle 6.2.400.76
򐂰 Brocade switch type 44, Fabric OS firmware version v6.3.0c
In this scenario, the XIV uses Microsoft Native MPIO and DS8000 LUNs managed by
SDDDSM device drivers.
Important: Initially installing SDDDSM on the host server might require an outage if a
reboot is needed.
The overall migration scenario is depicted in Figure 8-1. The three existing XIV data volumes
(shown on the left side of the diagram in solid lines) are configured on the Windows server
and are running live applications. The objective is to introduce the DS8000 into the
environment and migrate the XIV volumes (source volumes) to the DS8000 volumes (target
volumes). Three DS8000 volumes of equal or greater capacity using type ds are created in
DS8000. The target volume capacities (sizes) are shown on the right side of the diagram.
IBM XIV
IBM DS8100
To
ta lS to
r ga e
Windows Host Server

32 GB 40 GB
1 2
P7 2
48 GB 50 GB
64 GB 70 GB
FC Switch FC Switch
Figure 8-1 Migration test environment using Windows 2008 LDM
8.3.2 High-level plan for migration using Windows LDM mirroring

The high-level steps needed to migrate using Windows LDM mirroring are shown in Table 8-1.
Table 8-1 Example migration procedure using Windows LDM

Process Commands Explanation
Step 1. Identify the source 1. Right-click My Computer. This identification is done on

(XIV) LUNs on the host server 2. Click Manage. the Windows host server.
to be migrated. 3. Click Disk Management.

Step 2. Assign the DS8000 1. chvolgrp -dev image_id Step 1 is done on the DS8000
volumes to the Windows host -action add -volume vol_# with the dscli. You can choose
server and discover the new V# to use the GUI. The image_id is
volumes. 2. Click My Computer. the DS8000 id, and vol_num is
3. Click Manage. the number of the DS8000
4. Click Disk Management. volume in hex.
5. Click Action from the
toolbar. Steps 2-6 are performed on the
6. Click Rescan disks. Windows host server.
Step 3. Bring the DS8000 1. Right-click a new disk This step is done on the
target disks online. 2. Click Online. Windows host server.
Step 4. Identify the DS8000 1. Click Start This step is necessary if you
LUN IDs. 2. Select Subsystem Device have specific performance
DSM considerations. It is also
3. Click Subsystem Device needed if you have many LUNs
DSM that are the same size and you
4. Enter the command want to get LUN mapping
datapath query device details back to the DS8000
from the command prompt arrays.
Step 5. Initiate the mirroring 1. Right-click the source The synchronization of the
process. volume (x:). source and target volumes is
2. Click Add Mirror. automatic.
3. Select the chosen disk.
4. Click Add Mirror.
Step 6. Verify that the volumes Visual check Look for a disk that is
are copied and synced. functioning correctly.
Step 7. Remove the mirror. 1. Right-click the source disk. Make sure that you are
2. Click Remove Mirror. removing the source disk from
3. Select the source disk to the mirror, not the DS8000
remove. target disk.
4. Click Remove Mirror.
5. Click Yes.
6. Right-click the selected
volume.
8. Click Yes.
Step 8. Verify that the mirror is Visual check Check that the disk is now
removed. deallocated.
Step 9. Change the name of the 1. Right-click the volume. Change the name of the new
new target disk to IDBM 2107. 2. Click Properties. target disk to something
3. Enter the volume name meaningful.
under the General tab.
Step 10. Repeat steps 5-9. N/A Repeat the mirroring and
renaming process for the
remaining source and target
disks.
Step 11. Expand the DS8000 1. Right-click the volume. In the example, the target disks
disks to their full capacity. 2. Click Extend Volume. are larger than the source
3. Click the target disk. LUNs, so you must expand
4. Click Next. them to use the remaining
5. Click Finish. capacity.

Step 12. Delete the source 1. Unassign the volume in the Follow the procedures for each
volumes from the Windows host XIV box from the Windows component listed in the right
server. host server. column.
2. Delete the zone in the
fabric.
3. Disconnect the XIV storage
device from the fabric or
host server.
Step 13. Rescan the Windows 1. Right-click My Computer. If a volume is displayed as

host server. 2. Click Manage. missing, right-click the volumes
3. Click Disk Management. and click Delete.
4. Click Action from the
toolbar.
5. Click Rescan disks.
Step 14. Verify that the device Visual check Check that source volumes are
definitions are removed. gone.
8.3.3 Detailed steps for migration using Windows LDM mirroring

The procedure for doing the following steps is supplied for both the GUI and command line
when possible.
Identifying the source volumes on the Windows host server

To identify the source volumes on the host server, perform the following steps:
1. From the Windows 2008 desktop on the host server, click the Server Manager icon.
2. In the Server Manager window, click Storage and then Disk Management as shown in
Figure 8-2 on page 272.
Tip: You can also start Disk Management from the command line by typing
diskmgmt.msc.

Figure 8-2 View of the XIV source volumes on the Windows host server
3. Right-click the disk and select Properties.

4. Identify the disk manufacturer in the Properties window. In the example in Figure 8-2, the
three volumes are Disk 2, 3, and 4. The drive letters x:, y:, and z: are associated with basic
disks, which are the current source XIV volumes attached to the host.
Tip: To use the LDM function to mirror the LUNs, both the source and target disks need
to be dynamic disks. For Windows 2008, you do not have to manually convert to
dynamic disks because it is part of the mirroring process.

Assigning the DS8000 volumes to the Windows host server
Using the DS8000 dscli or DS GUI, assign the target LUNs to the host server using the
following steps. Make sure that the target LUNs are the same size or larger than the source
LUNs.
Important: Before proceeding, verify that you have properly cabled and zoned the
connections from the DS8000 to the host server through the SAN fabric.
1. Assign the volumes, use the dscli command as shown in Example 8-1.
Example 8-1 Using the dscli commands to assign storage to the Windows host
mkfbvol -dev IBM.2107-7581981 -extpool P2 -name 8300_#h -type ds -cap 40 2000
mkvolgrp -dev IBM.2107-75L4741 -type scsimap256 -volume 2000-2002 V6
mkhostconnect -dev IBM.2107-7581981 -wwname 210000E08B875833 -hosttype Win2008
-volgrp V6 x345-tic-17
mkhostconnect -dev IBM.2107-7581981 -wwname 210000E08B09E5FD -hosttype Win2008
-volgrp V6 x345-tic-17
2. Verify that the LUNs are indeed assigned to the volume group V6 by issuing the
showvolgrp command as shown in Example 8-2.
Example 8-2 Using the showvolgrp command to verify the volumes in a volgrp from the dscli
dscli> showvolgrp -dev IBM.2107-75L4741 V6
Date/Time: March 28, 2007 1:18:43 PM CST IBM DSCLI Version: 5.2.400.304 DS:
IBM.2107-75L4741
Name V6
ID V6
Type SCSI Map 256
Vols 2000 2001 2002

Windows 2008 automatically discovers the target LUNs. If the LUNs do not display on the
Disk Management window, click Action  Rescan as shown in Figure 8-3.
Figure 8-3 Running the Rescan tool on Windows

The discovered LUNs are displayed as shown in Figure 8-4.
Figure 8-4 View of disks after the Rescan
In this example, Disks 4, 5, and 6 are the new DS8000 disks that will be part of the mirror
(the target volumes). The disks are offline and not allocated.
3. Use command-line utility diskpart to rescan the disks as shown in Example 8-3.
Example 8-3 Rescan disk using diskpart
DISKPART> rescan
DISKPART> list disk
Bringing the DS8000 target disks online

Microsoft Windows 2008 introduced SAN policy to protect shared disks accessed by multiple
servers. The default policy for Windows Advanced Server and Windows Data Center is
OfflineShared. In this case, the boot disk and all disks that are not located on a shared bus

such as SCSI, iSCSI, or SAS is are brought online. The offline disks are read-only by default.
On all other versions of Windows, the default is to bring all disks online. In this case, the disks
are online and read/write.
You can use Diskpart utility to set the POLICY=OnlineAll to get around the default policy.
However, if the disks are shared among servers, be aware that this setting can lead to data
corruption. Use the correct SAN policy to protect your data.
After being brought online once, offline disks will be automatically online after the reboot.
To bring the target disks online, right-click one of the new disks and select Online as shown in
Figure 8-5.
Figure 8-5 Bringing disks online
You can also use command-line utility Diskpart to make the disks online (Example 8-4).
Example 8-4 Bringing a disk online with diskpart
DISKPART> select disk 4

DISKPART> attributes disk clear readonly
DISKPART> online disk
Identifying specific source and target LUNs

If you have several disks of the same size and want to mirror specific volumes, you must
identify the DS8000 volumes by the DS8000 LUN ID.

You can identify the LUNs by using the SDD command-line interface (SDDDSM) using the
following steps:
1. From your Windows desktop, click Start.
2. Select Subsystem Device DSM  Subsystem Device Driver DSM (Figure 8-6).
Figure 8-6 Using the SDDDSM CLI
3. In the DSMCLI window, enter the command datapath query device.

The LUNs are shown in the red circles. The LUN IDs are represented by the last four digits
of the serial numbers. For example, LUN 2000 at the top of the window is from LSS 20 and
is the first LUN (00) from that LSS. The naming conventions were set so you know that
LSS 20 is made from extent pool 2, and array 2 is located in extpool 2. Because each LUN
is showing up through two paths, the SDD driver assigns one Dev number per two paths.
For example, LUN 2000 is DEV 0, Disk 4.
Note: If you have specific requirements to share or isolate applications and files on
separate arrays, map each LUN back to the DS8000. Spread or isolate these LUNs
during the mirroring process.

Initiating the mirroring process
Now, with Disk 4, 5, and 6 brought online, the system is ready to initiate the mirroring process.
To initiate the process, perform the following steps:
1. From the Disk Management window, right-click the source volume (x:) and select Add
Mirror (Figure 8-7).
Figure 8-7 Accessing Add Mirror window

2. The Add Mirror window displays a list of available disks. Select the disk and click Add
Mirror as shown in Figure 8-8.
Figure 8-8 Selecting disks in the Add Mirror window
3. In this example, the original disk is still a basic disk. A warning opens to confirm that you
want to convert it to a dynamic disk as shown in Figure 8-9. Click Yes to continue.
Figure 8-9 Warning of converting to dynamic disk

After the mirror is added, the synchronization process starts automatically. At this time,
you can see that both volumes, Disk 1 and Disk 4, are assigned to the same drive letter, x:
In Figure 8-10, the sync process is at 6%.
Figure 8-10 Synchronization process running
You can also use command-line utility Diskpart to start the mirroring as shown in
Example 8-5.
Example 8-5 Initiate the mirroring process with diskpart

DISKPART> convert dynamic
DISKPART> convert dynamic
DISKPART> add disk=4

Verifying that the volumes are copied and synchronized
Figure 8-11 shows the volumes after the synchronization process completes. Notice that the
new volume is now healthy.
Figure 8-11 Synchronization process finished
You can also check the volume synchronization process with Diskpart as shown in
Example 8-6. Notice that Volume 3 is rebuilding.
Example 8-6 Show the volume status with Diskpart


---------- --- ----------- ----- ---------- ------- --------- --------
Volume 0 X XIV NTFS Simple 40 GB Healthy
Volume 2 D LAB-DATA FAT32 Partition 2000 MB Healthy
* Volume 3 Y XIV NTFS Mirror 48 GB Rebuild
Volume 4 Z XIV NTFS Partition 64 GB Healthy

Removing the mirror
In this step, you must choose one of two options:
򐂰 Break the mirrored volume: The selected volume keeps the original drive letter, and the
other volume is automatically assigned another letter. The synchronization process no
longer occurs. However, the data is retained.
򐂰 Remove the mirror: If you choose to remove the mirror, a window is displayed that asks
you which volume you want to remove. The selected volume becomes a free disk with no
drive letter assigned, and all data on it is erased.
In the example, the XIV source disk is no longer wanted in the environment, so you remove
the mirror. To remove the mirror, perform these steps:
1. Right-click the disk that you want to remove the mirror and select Remove Mirror
(Figure 8-12).
Figure 8-12 Remove Mirror

2. In the Remove Mirror window, select the source XIV disk to remove as shown in
Figure 8-13. In the example, Disk 1 is selected because it is the XIV source volume.
Figure 8-13 Remove mirror, continued

4. Click Yes to confirm the removal.
You can also use Diskpart to remove the mirror as shown in Example 8-7.
Example 8-7 Remove mirror with diskpart
DISKPART> break disk=5 nokeep

DiskPart successfully broke the mirror volume.

Verifying that the mirror is removed
Verify that the selected volume is now available to the operating system without a drive letter
or data as shown in Figure 8-14.
Figure 8-14 Verifying removing mirror window
To verify that the mirror is removed with Diskpart, list the volume and check if the type is
simple. If the mirror is not removed, the volume is listed as mirror as shown in Example 8-8.
Example 8-8 Verifying that the mirror removed with diskpart

---------- --- ----------- ----- ---------- ------- --------- --------
Volume 0 X IBM 2107 NTFS Simple 40 GB Healthy
Volume 2 D LAB-DATA FAT32 Partition 2000 MB Healthy
Volume 3 Y XIV NTFS Simple 48 GB Healthy
Volume 4 Z XIV NTFS Partition 64 GB Healthy

Changing the name of the new target disk
Change the name of the new target disk to something meaningful using the following steps:
1. Right-click the volume and select Properties.
2. Name the volume under the General tab as shown in Figure 8-15. In the example, the
volume is called an IBM 2107 volume.
Figure 8-15 Changing the volume properties
Repeating the steps for the remaining disks

Repeat the steps starting with “Initiating the mirroring process” on page 278 for the remaining
disks. You can mirror multiple disks in parallel.

Extending the target volumes
If you select target volumes that are larger than the source volumes, you need to extend the
volumes to use the unused capacity on the target disks. To do extend the volumes, perform
1. Right-click the volume and select Extend Volume as shown in Figure 8-16.
Figure 8-16 View of extending the volumes to full capacity

2. In the Extend Volume Wizard window, right-click the disk you want to extend and select
Extend Volume (Figure 8-17).
Figure 8-17 View of the Extend Volume Wizard window
3. Click Next.
4. Click Finish.
The volume is now extended as shown in Figure 8-18.
Figure 8-18 View of extended volume
You can also extend the volume with Diskpart as shown in Example 8-9.
Example 8-9 Extend volume with diskpart

Volume 1 is the selected volume.
DISKPART> extend
DiskPart successfully extended the volume.

Deleting the source volumes from the Windows host
After the data is migrated, you need to delete the original volumes. To do so, perform the
following steps:
1. Unassign the volume in the source storage subsystem (XIV) from the Windows host
server.
2. Delete the zone in the fabric.
3. Uncable (disconnect) the source (XIV) storage subsystem from the fabric or host server.
After this process is complete, visually verify that the device definitions are removed. If the
volumes are displayed as missing, right-click the volumes and click Delete.
8.4 Data migration using Solaris Volume Manager

This section addresses the tasks required to migrate from two existing EMC LUNs to two new
DS8000 LUNs using the Solaris Volume Manager (SVM) mirroring capabilities.
The process addressed here is fairly generic. The number of LUNs and the type of storage
subsystem they originate from do not really matter. The process is the same for two or twenty
LUNs from any subsystem. Depending on the specifics of your environment, the process can
be disruptive to applications, requiring reboots and file system unmounts.
The example environment has the following characteristics:

򐂰 The data to be migrated is on two EMC LUNs.
򐂰 Each LUN has one logical volume that is on slice 2. Using slice 2 in Solaris allows you to
cover the entire disk.
򐂰 Disk numbers 2 and 3 need to be migrated (Figure 8-19 on page 289) to DS8000 storage.
򐂰 The source volumes to be migrated are already under control of SVM and coded
appropriately in /etc/vfstab.
򐂰 The source volumes are not SVM submirrors to an existing mirror.
򐂰 Because ensuring that the target LUNs are the same size as the source LUNs is difficult,
the DS8000 LUNs are a little larger than the source LUNs.
򐂰 The same host system (Solaris) is attached to both the source and the destination storage
subsystems.
8.4.1 Test environment

The test environment was composed of the following components:
򐂰 Sun Fire V280R host system
– Solaris 10 11/06 (s10s_u3wos_10 SPARC) with Solaris Fibre Channel (FC) and
Storage Multipathing software
– Sun badged QLogic-based QLA2340 HBAs (SG-XPCI1FC-QL2) running firmware
version 3.3.117
򐂰 EMC - 8730-18 - EMC Symmetrix 5 Storage running microcode level 5568.68

Figure 8-19 shows the hardware layout.
EMC Symmetrix 8730 IBM DS8100
Enterpri se Sto rage System
TotalStorage
Sun Fire V280R Server

33.71 GB 33.98 GB
1 2
8.43 GB 9.00 GB
Figure 8-19 Hardware layout
8.4.2 Solaris Volume Manager mirroring process

The Solaris Volume Manager mirroring process consists of the following basic steps:
1. Configure Solaris FC and multipathing software to recognize DS8000 storage: In this
case, the server is equipped with Sun-badged HBAs. Sun HBAs allow you to use the Sun
multipathing solution (Solaris Fibre Channel (FC) and Storage Multipathing software).
SVM, other than Symantec VxVM, does not provide any built-in multipathing capability.
2. Determine source LUN information: This step needs to be done to discover which SVM
object is associated with which device special file. You also need to know the size of the
source LUNs to create appropriately sized target LUNs.
3. Discover the DS800 target LUNs.
4. Make the source volume a mirror member: This step might not be necessary if the source
volumes are already a submirror to an existing mirror.
5. Make the source LUN SVM objects.
6. Make the destination volume a mirror member and synchronize the mirror: During this
step, the source data is copied to the target devices.
7. Remove the source volumes from the mirror: This step removes the source LUNs from the
SVM structures.

8.4.3 High-level plan for migration using SVM mirroring
The high-level steps are shown in Table 8-2.
Table 8-2 Example migration procedure using Solaris SVM
Step 1. Configure Solaris Fibre vi Open a vi session and update

Channel (FC) and Storage /kernel/drv/scsi_vhci.conf that file as shown in the detailed
Multipathing software to section.
recognize DS8000 storage.
Step 2. Reboot the system. stmsboot -u Reboot to activate this change

and enable multipathing for the
DS8000 LUNs.
Step 3. Determine information 1. more /etc/vfstab Determine the association of

about the migration source 2. metastat d21 the file system or mount points
LUNs. that are on the source volume
with the underlying SVM object.
You can determine the
association by reviewing the
contents of /etc/vsfstaband
using the metastat command.
Step 4. Discover the DS8000 format Determine which LUNs belong

migration destination LUNs. to the DS8000.
Step 5. Make the source 1. metainit This step is disruptive to the

volume a mirror member. 2. vi /etc/vfstab application because it requires
3. df -k the migration source volumes to
be unmounted. If the source
LUNs are already submirrors,
this step can be omitted.
Run the command metaintit.
Modify the vsfstab.
Mount the mirrors.
Verify the mount with df -k.
Step 6. Make the source LUNs 1. metainit d22 1 1 The next command sequence
SVM objects. c5t6005076305FFC78600000 brings the DS8000 migration
00000004001d0s2 destination volumes under the
2. metainit d32 1 1 control of Solaris SVM. They
c5t6005076305FFC78600000 are designated as d22 and d32.
00000004003d0s
3. metastat
Step 7. Make the destination 1. metattach d20 d22 Create a two-way mirror by
volume a mirror member and 2. metattach d30 d32 integrating the metadevice
sync the mirror. (d22) based on DS8000 into the
already existing mirror (d20).
Step 8. Remove the source 1. metadetach d20 d21 After the sync is complete,
volumes from the mirror. 2. metastat d20 remove the mirror and verify
3. metadetach d30 d31 with the metastat command.
4. metastat d30

8.4.4 Detailed steps for migration using SVM mirroring
The migration procedure has the following detailed steps:
򐂰 Configuring Solaris FC and multipathing software for DS8000
򐂰 Rebooting the system
򐂰 Determining source LUN information
򐂰 Discovering the DS8000 target LUNs
򐂰 Making the source volume a mirror member
򐂰 Making the source LUNs SVM objects
򐂰 Making the destination volume a mirror member and synchronize
򐂰 Removing source volumes from the mirror and verifying the migration
Configuring Solaris FC and multipathing software for DS8000

To introduce a third-party symmetric disk subsystem to Solaris Fibre Channel (FC) and
Storage Multipathing software, supply this subsystem information in the file scsi_vhci.conf:
򐂰 The vendor identification (VID)
򐂰 The product identification (PID) string
This file is in the /kernel/drv directory. The file needs to be updated as shown in Figure 8-20.
# vi /kernel/drv/scsi_vhci.conf
"/kernel/drv/scsi_vhci.conf" 31 lines, 1053 characters
#
# Copyright 2004 Sun Microsystems, Inc. All rights reserved.
# Use is subject to license terms.
#
#pragma ident "@(#)scsi_vhci.conf 1.9 04/08/26 SMI"
#
name="scsi_vhci" class="root";
#
load-balance="round-robin";
#
auto-failback="enable";
#
# For enabling MPxIO support for 3rd party symmetric device need an
# entry similar to following in this file. Just replace the "SUN SENA"
# part with the Vendor ID/Product ID for the device, exactly as reported by
# Inquiry cmd.
#
# device-type-scsi-options-list =
device-type-scsi-options-list =
# "SUN SENA", "symmetric-option";
"EMC SYMMETRIX", "symmetric-option",
"IBM 2107900", "symmetric-option";
# symmetric-option = 0x1000000;
symmetric-option = 0x1000000;
Figure 8-20 Enable DS8000 for Solaris multipath support

Rebooting the system
The system needs to be rebooted to activate the changes and enable multipathing for the
DS8000 LUNs. Issue the command stmsboot -u as shown in Figure 8-21.
# stmsboot -u
WARNING: This operation will require a reboot.
Do you want to continue ? [y/n] (default: y) y
The changes will come into effect after rebooting the system.
Reboot the system now ? [y/n] (default: y) y# stmsboot -u
WARNING: This operation will require a reboot.
Do you want to continue ? [y/n] (default: y) y
Figure 8-21 Using stmsboot to enable multipathing for the DS8000 LUNs
Determining source LUN information

You need to gather information and understand several aspects about the migration source
volumes and file systems. To do so, perform the following steps:
1. Determine the association of the file systems or mount points that are on the source
volume with the underlying SVM object. Review the contents of the /etc/vfstab file and
display the mounted file systems using the df command as shown in Example 8-10.
In this example, the file system on metadevice /dev/md/dsk/d21 is mounted on mount
point /emc08. The file system on metadevice /dev/md/dsk/d31 is mounted on mount point
/emc19.
Example 8-10 \Showing the association between mount points or file systems, and an SVM object
# more /etc/vfstab
#device device mount FS fsck mount mount
#to mount to fsck point type pass at boot options
#
fd - /dev/fd fd - no -
/proc - /proc proc - no -
# non-mpxio: /dev/dsk/c2t0d0s1 - - swap - no -
# mpxio: /dev/dsk/c5t20000004CFA3C467d0s1 - - swap - no -
/dev/dsk/c5t20000004CFA3C467d0s1 - - swap - no -
# non-mpxio: /dev/dsk/c2t0d0s0 /dev/rdsk/c2t0d0s0 / ufs 1 no -
# mpxio: /dev/dsk/c5t20000004CFA3C467d0s0 /dev/rdsk/c5t20000004CFA3C467d0s0 /
ufs 1 no -
/dev/dsk/c5t20000004CFA3C467d0s0 /dev/rdsk/c5t20000004CFA3C467d0s0 / ufs
1 no -
# non-mpxio: /dev/dsk/c2t0d0s7 /dev/rdsk/c2t0d0s7 /export/home ufs 2 yes
-
# mpxio: /dev/dsk/c5t20000004CFA3C467d0s7 /dev/rdsk/c5t20000004CFA3C467d0s7
/export/home ufs 2 yes -
/dev/dsk/c5t20000004CFA3C467d0s7 /dev/rdsk/c5t20000004CFA3C467d0s7
/dev/md/dsk/d21 /dev/md/rdsk/d21 /emc08 ufs 2 yes -
/devices - /devices devfs - no -
ctfs - /system/contract ctfs - no -
objfs - /system/object objfs - no -
swap - /tmp tmpfs - yes -
#
#
# df
/ (/dev/dsk/c5t20000004CFA3C467d0s0): 3361778 blocks 482685 files

/devices (/devices ): 0 blocks 0 files
/system/contract (ctfs ): 0 blocks 2147483614 files
/proc (proc ): 0 blocks 29948 files
/etc/mnttab (mnttab ): 0 blocks 0 files
/etc/svc/volatile (swap ): 4223920 blocks 295309 files
/system/object (objfs ): 0 blocks 2147483452 files
/platform/sun4u-us3/lib/libc_psr.so.1(/platform/sun4u-us3/lib/libc_psr/libc_psr_hwcap1.so.1
): 3361778 blocks 482685 files
/platform/sun4u-us3/lib/sparcv9/libc_psr.so.1(/platform/sun4u-us3/lib/sparcv9/libc_psr/libc
_psr_hwcap1.so.1): 3361778 blocks 482685 files
/dev/fd (fd ): 0 blocks 0 files
/tmp (swap ): 4223920 blocks 295309 files
/var/run (swap ): 4223920 blocks 295309 files
/export/home (/dev/dsk/c5t20000004CFA3C467d0s7):58488562 blocks 3523516 files
/emc08 (/dev/md/dsk/d21 ):14855432 blocks 1048314 files
/emc19 (/dev/md/dsk/d31 ):67023108 blocks 4193274 files
2. Discover the underlying LUNs of the metadevices by running the metastat command as
shown in Example 8-11. In this example, the Solaris disk device
/dev/dsk/c5t6006048000028470097553594D304342d0s2 is associated with the SVM
metadevice d21.
Example 8-11 Issuing the metastat command

# metastat d21
d21: Concat/Stripe
Size: 17671680 blocks (8.4 GB)
Stripe 0:
Device Start Block Dbase Reloc
/dev/dsk/c5t6006048000028470097553594D304342d0s2 0 No Yes
Device Relocation Information:

Device Reloc Device ID
/dev/dsk/c5t6006048000028470097553594D304342d0 Yes
id1,ssd@n6006048000028470097553594d304342
#
#
# metastat d31
d31: Concat/Stripe
Size: 70690560 blocks (33 GB)
Stripe 0:
/dev/dsk/c5t6006048000028470097553594D304533d0s2 0 No Yes

/dev/dsk/c5t6006048000028470097553594D304533d0 Yes
id1,ssd@n6006048000028470097553594d304533
3. Determine the size of the migration source LUNs so you know what size is required for the
target LUN using the format command (Example 8-12).
Example 8-12 Output of the format command

# format
c5t6006048000028470097556434D303030d0: configured with capacity of 6.56MB


/pci@8,600000/SUNW,qlc@4/fp@0,0/ssd@w500000e0102fee11,0
1. c5t20000004CFA3C467d0 <SUN36G cyl 24620 alt 2 hd 27 sec 107>
/scsi_vhci/ssd@g20000004cfa3c467
2. c5t6006048000028470097556434D303030d0 <EMC-SYMMETRIX-5568 cyl 14 alt
2 hd 15 sec 64>
/scsi_vhci/ssd@g6006048000028470097556434d303030
3. c5t6006048000028470097553594D304342d0 <EMC-SYMMETRIX-5568 cyl 18408
alt 2 hd 15 sec 64>
/scsi_vhci/ssd@g6006048000028470097553594d304342
alt 2 hd 30 sec 64>
/scsi_vhci/ssd@g6006048000028470097553594d304533
alt 2 hd 15 sec 64>
/scsi_vhci/ssd@g6006048000028470097553594d304638
In the sample environment, one unformatted EMC LUN with a device special file of
c5t6006048000028470097556434D303030d0 and a size of 6.56 MB exists. It is the
Volume Configuration Management Database LUN (VCMDB) that is used to store host
attachment-specific information. It must not be used to store data.
4. Select the disk you want to migrate in the format menu (Example 8-13).
Example 8-13 Using the format dialog

# format

c5t6005076305FFC7860000000000004001d0: configured with capacity of 9.00GB

2. c5t6006048000028470097556434D303030d0 <EMC-SYMMETRIX-5568 cyl 14 alt
2 hd 15 sec 64>
/scsi_vhci/ssd@g6006048000028470097556434d303030
alt 2 hd 15 sec 64>
/scsi_vhci/ssd@g6006048000028470097553594d304342
alt 2 hd 30 sec 64>
/scsi_vhci/ssd@g6006048000028470097553594d304533
alt 2 hd 15 sec 64>
/scsi_vhci/ssd@g6006048000028470097553594d304638
6. c5t6005076305FFC7860000000000004001d0 <IBM-2107900-.437 cyl 9828 alt
2 hd 30 sec 64>
/scsi_vhci/ssd@g6005076305ffc7860000000000004001
2 hd 30 sec 64>

2 hd 30 sec 64>
2 hd 64 sec 256>
5. Issue the partition command to print the partition table (Example 8-14).
Example 8-14 Using the partition dialog

selecting c5t6006048000028470097553594D304342d0
[disk formatted]
FORMAT MENU:
disk - select a disk
type - select (define) a disk type
partition - select (define) a partition table
current - describe the current disk
format - format and analyze the disk
repair - repair a defective sector
label - write label to the disk
analyze - surface analysis
defect - defect list management
backup - search for backup labels
verify - read and display labels
save - save new disk/partition definitions
inquiry - show vendor, product and revision
volname - set 8-character volume name
quit
format> pa
6. Issue the print command to print the partition information (Example 8-15). In the example
environment, slice 2 of disk 3 is 8.43 GB. A slice in Solaris is a partition on the disk.
Typically, slice 2 refers to the entire disk.
Example 8-15 Using the print dialog

format> pa
PARTITION MENU:
0 - change `0' partition
select - select a predefined table
modify - modify a predefined partition table
name - name the current table
print - display the current table
label - write partition map and label to the disk
quit

partition> pr
Current partition table (original):
Total disk cylinders available: 18408 + 2 (reserved cylinders)
Part Tag Flag Cylinders Size Blocks

0 root wm 0 - 273 128.44MB (274/0/0) 263040
1 swap wu 274 - 547 128.44MB (274/0/0) 263040
2 backup wu 0 - 18407 8.43GB (18408/0/0) 17671680
3 unassigned wm 0 0 (0/0/0) 0
6 usr wm 548 - 18407 8.18GB (17860/0/0) 17145600
Discovering the DS8000 target LUNs

In the example, two DS8000 migration destination LUNs were created on the DS8000. These
LUNs are slightly larger than the source LUNs. Creating the target LUNs that are slightly
larger is easier than trying to make them the same size. The DS8000 LUNs are one 9 GB
LUN and one 33.9 GB LUN.
Note: DS8000 volumes for Solaris must be configured with type = scsimap256 and volume
type = Sun.
Example 8-16 shows 5 disks/LUNs listed that are not labeled by Solaris. The DS8000 storage
administrator had assigned twice the number of LUNs as requested. The LUN sizes are listed
in the unlabeled section (first section).
Example 8-16 Output of the command format with DS8000 LUNs attached
# format

2. c5t6006048000028470097556434D303030d0 <EMC-SYMMETRIX-5568 cyl 14 alt 2 hd 15 sec
64>
/scsi_vhci/ssd@g6006048000028470097556434d303030
3. c5t6006048000028470097553594D304342d0 <EMC-SYMMETRIX-5568 cyl 18408 alt 2 hd 15
sec 64>
/scsi_vhci/ssd@g6006048000028470097553594d304342
sec 64>
/scsi_vhci/ssd@g6006048000028470097553594d304533
sec 64>
/scsi_vhci/ssd@g6006048000028470097553594d304638
6. c5t6005076305FFC7860000000000004001d0 <IBM-2107900-.437 cyl 9828 alt 2 hd 30 sec
64>

64>
64>
256>
selecting c5t6005076305FFC7860000000000004001d0
[disk formatted]
The first LUN is the EMC storage that was noted previously. By comparing the device special
file number in the unlabeled LUNs and the numbered LUNs shown in the AVAILABLE DISK
SELECTIONS: section, you can determine which LUNs belong to the DS8000. See the disk
list in the format command output in Example 8-13 on page 294.
Select the LUNs that you want to use as migration targets as shown at the bottom of
Example 8-16 on page 296. In the example, LUNs 6 and 9 are chosen as the DS8000
migration targets.
Table 8-3 summarizes the source and target LUNs used in the scenario. It shows the native
device special file names, the associated metadevice names, and the respective sizes.
Table 8-3 Showing the source and target volumes

Source Size of source (GB) Target Size of target (GB)
c5t600604800002847 8.43 c5t6005076305FFC78 9.00

0097553594D304342d 60000000000004001d
0 0
d21 d22
c5t600604800002847 33.71 c5t6005076305FFC78 33.98

0097553594D304533d 60000000000004003d
0 0
d31 d32
Making the source volume a mirror member
Important: This step is disruptive to the application because it requires the migration
source volumes to be unmounted. If the source LUNs are already submirrors, this step can
be omitted.
The command sequence shown in Example 8-17 includes the following steps:
1. Create a one-way mirror using the primary volume names (d20 and d30 in the example)
2. Make the source volumes (d21 and d31) submirrors.
3. Unmount the file systems on the disks that are to be migrated. You will remount them later,
using the name of the mirror rather than the name of the submirror.
Example 8-17 Creating a one-way mirror and taking DS8000 LUNs under control of SVM
# metainit d20 -m d21
d20: Mirror is setup
#
# metainit d30 -m d31

d30: Mirror is setup
#
# umount /emc08
# umount /emc19
#
Edit /etc/vfstab to reflect the new primary volume (d20 and d30) names. Example 8-18
shows what /etc/vfstab looks like after these changes are completed.
Example 8-18 Mount the mirrors

# vi /etc/vfstab
"/etc/vfstab" 20 lines, 1036 characters
#device device mount FS fsck mount mount
#to mount to fsck point type pass at boot options
#
fd - /dev/fd fd - no -
/proc - /proc proc - no -
# non-mpxio: /dev/dsk/c2t0d0s1 - - swap - no -
# mpxio: /dev/dsk/c5t20000004CFA3C467d0s1 - - swap - no -
/dev/dsk/c5t20000004CFA3C467d0s1 - - swap - no -
# non-mpxio: /dev/dsk/c2t0d0s0 /dev/rdsk/c2t0d0s0 / ufs 1 no -
# mpxio: /dev/dsk/c5t20000004CFA3C467d0s0 /dev/rdsk/c5t20000004CFA3C467d0s0 /
ufs 1 no -
/dev/dsk/c5t20000004CFA3C467d0s0 /dev/rdsk/c5t20000004CFA3C467d0s0 / ufs
1 no -
# non-mpxio: /dev/dsk/c2t0d0s7 /dev/rdsk/c2t0d0s7 /export/home ufs 2 yes
-
# mpxio: /dev/dsk/c5t20000004CFA3C467d0s7 /dev/rdsk/c5t20000004CFA3C467d0s7
/dev/dsk/c5t20000004CFA3C467d0s7 /dev/rdsk/c5t20000004CFA3C467d0s7
/devices - /devices devfs - no -
ctfs - /system/contract ctfs - no -
objfs - /system/object objfs - no -
swap - /tmp tmpfs - yes -
#

Remount the file systems and verify that they are mounted with the df command as shown in
Example 8-19.
Example 8-19 Display file systems

df -k
...
/dev/md/dsk/d30 34809572 1298018 33163459 4% /emc19
/dev/md/dsk/d20 8701901 7837153 777729 91% /emc08
Making the source LUNs SVM objects

The next command sequence (Example 8-20) brings the DS8000 migration destination
volumes under the control of Solaris SVM, They are designated as d22 and d32.
Example 8-20 Making SVM objects out of the target LUNs

# metainit d22 1 1 c5t6005076305FFC7860000000000004001d0s2
d22: Concat/Stripe is setup
#
# metainit d32 1 1 c5t6005076305FFC7860000000000004003d0s2
d32: Concat/Stripe is setup
#
Use the metastat command to show the hierarchal relationship between the primary volume
and its submirror members as shown in Example 8-21. The metadevices created on the
DS8000 LUNs (d22 and d32) do not belong to a mirror.
Example 8-21 Using the command metatat to show the status of the mirror
# metastat
d30: Mirror
Submirror 0: d31
State: Okay
Pass: 1
Read option: roundrobin (default)
Write option: parallel (default)
d31: Submirror of d30
State: Okay
Stripe 0:
Device Start Block Dbase State
Reloc Hot Spare
/dev/dsk/c5t6006048000028470097553594D304533d0s2 0 No Okay
Yes
d20: Mirror
Submirror 0: d21
State: Okay
Pass: 1
State: Okay
Stripe 0:

Reloc Hot Spare
Yes
d32: Concat/Stripe
Stripe 0:
/dev/dsk/c5t6005076305FFC7860000000000004003d0s2 0 No Yes
d22: Concat/Stripe
Stripe 0:
/dev/dsk/c5t6005076305FFC7860000000000004001d0s2 0 No Yes
/dev/dsk/c5t6005076305FFC7860000000000004003d0 Yes
id1,ssd@n6005076305ffc7860000000000004003
id1,ssd@n6005076305ffc7860000000000004001
/dev/dsk/c5t6006048000028470097553594D304533d0 Yes
id1,ssd@n6006048000028470097553594d304533
/dev/dsk/c5t6006048000028470097553594D304342d0 Yes
id1,ssd@n6006048000028470097553594d304342
Making the destination volume a mirror member and synchronize

Create a two-way mirror by integrating the metadevice (d22) based on DS8000 into the
already existing mirror (d20) using the metattach command. This command starts the
synchronization process shown in Example 8-22. You can check the progress of the mirror
synchronization using the metastat command.
Example 8-22 Using metattach to create a two-way mirror and check the status
# metattach d20 d22
d20: submirror d22 is attached
#
#
# metastat d20
d20: Mirror
Submirror 0: d21
State: Okay
Submirror 1: d22
State: Resyncing
Resync in progress: 3 % done
Pass: 1

State: Okay
Stripe 0:
Device Start Block Dbase
State Reloc Hot Spare
/dev/dsk/c5t6006048000028470097553594D304342d0s2 0 No
Okay Yes

State: Resyncing
Stripe 0:
/dev/dsk/c5t6005076305FFC7860000000000004001d0s2 0 No
Okay Yes

/dev/dsk/c5t6006048000028470097553594D304342d0 Yes
id1,ssd@n6006048000028470097553594d304342
id1,ssd@n6005076305ffc7860000000000004001
##
Repeat the steps for the other mirror (d30) and submirror (d32) as shown in Example 8-23.
Example 8-23 Using metattach to create a two-way mirror and check the status
metattach d30 d32
d30: submirror d32 is attached
#
#
# metastat d30
d30: Mirror
Submirror 0: d31
State: Okay
Submirror 1: d32
State: Resyncing
Resync in progress: 3 % done
Pass: 1

State: Okay
Stripe 0:
Reloc Hot Spare
Yes

State: Resyncing
Stripe 0:
Reloc Hot Spare
/dev/dsk/c5t6005076305FFC7860000000000004003d0s2 0 No Okay
Yes

/dev/dsk/c5t6006048000028470097553594D304533d0 Yes
id1,ssd@n6006048000028470097553594d304533
id1,ssd@n6005076305ffc7860000000000004003
Monitor the state of the destination volume submirror until the status is Okay. This status
indicates that the mirror is in sync and complete as seen in Example 8-24.
Example 8-24 Mirror synchronization is complete

# metastat d20
d20: Mirror
Submirror 0: d21
State: Okay
Submirror 1: d22
State: Okay
Pass: 1

State: Okay
Stripe 0:
/dev/dsk/c5t6006048000028470097553594D304342d0s2 0 No
Okay Yes

State: Okay
Stripe 0:
/dev/dsk/c5t6005076305FFC7860000000000004001d0s2 0 No
Okay Yes

The current configuration is shown in Figure 8-22.
d20 Mirror d30 Mirror
d21 Submirror d22 Submirror d31 Submirror d32 Submirror
EMC Source
EMC Source /dev/dsk/c5t6006048000028
dev/dsk/c5t6006048000028 470097553594D304533d0s2
470097553594D304342d0s2
I Source
IBM Target dev/dsk/c5t6005076305FFC786
dev/dsk/c5t6005076305FFC7 0000000000004003d0s2
860000000000004001d0s2
Figure 8-22 Entire setup after full mirroring is established
Removing source volumes from the mirror and verifying the migration
The two-way mirror is now synchronized, so you can remove the original source volumes from
the SVM. Use the metadetach command to detach the mirrors and submirrors as shown in
Example 8-25. In the example, detach mirror d20 and submirror d21, and mirror d30 and
submirror d31.You can use the metastat command to verify the results.
Example 8-25 Using metadetach to remove a metaset from a mirror

# metadetach d20 d21
d20: submirror d21 is detached
# metastat d20
d20: Mirror
Submirror 1: d22
State: Okay
Pass: 1

State: Okay
Stripe 0:
Reloc Hot Spare
Yes
id1,ssd@n6005076305ffc7860000000000004001
#
# metadetach d30 d31
d30: submirror d31 is detached
# metastat d30
d30: Mirror
Submirror 1: d32

State: Okay
Pass: 1
State: Okay
Stripe 0:
Reloc Hot Spare
Yes

id1,ssd@n6005076305ffc7860000000000004003
The migration is now complete, and all data is located exclusively on DS8000 LUNs. There
might be further tasks remaining such as disconnecting the source LUNs from the system.
8.5 Data migration using Veritas Storage Foundation

This section addresses migrating data from three existing EMC LUNs to three DS8000 LUNs
using the Veritas Volume Manager (VxVM) mirroring capabilities. This description is fairly
generic. The number of LUNs does not really matter: The process is the same for three or
more.
The example in this section describes an environment with the following characteristics:
򐂰 The data you want to migrate is on three EMC LUNs. It can, however, be on any other type
of storage system without affecting the process described.
򐂰 Each source LUN has one logical volume on slice 2. Slice 2 in Solaris covers the entire
disk.
򐂰 The EMC source volumes are already under control of VxVM and coded appropriately in
/etc/vfstab. They are mounted on /emcN , 0 <= N <= 2.
򐂰 You are migrating the data from three source LUNs to three target LUNs on a DS8000. It is
difficult to ensure that the target LUNs are the same size as the source LUNs. Therefore,
use DS8000 LUNs are a little larger than the source LUNs.
򐂰 The same host system is attached to both the source and the destination storage
subsystems.

The physical test environment is composed of the following components:
򐂰 Sun Fire V280R host system
– Solaris 10 11/06 (s10s_u3wos_10 SPARC)
– Sun-badged QLogic based QLA2340 HBAs (SG-XPCI1FC-QL2) running firmware
version 3.3.117

– Veritas Storage Foundation, including Veritas File System (VxFS) 5.0 and Veritas
Volume Manager (VxVM) 5.0. When VxVM is installed, a subcomponent called
Dynamic Multipathing (DMP) is installed automatically.
򐂰 EMC - 8730-18 - EMC Symmetrix 5 Storage running microcode level 5568.68
Figure 8-23 gives an overview of the hardware layout.
EMC Symmetrix 8730 IBM DS8100
E nterprise S to rage System

Sun Fire V280R Server
T otalStorage
33.71 GB
9.00 GB
1 2
9.00 GB
8.43 GB
8.43 GB 9.00 GB
Figure 8-23 HW layout
8.5.2 Veritas Volume Manager mirroring

The Veritas Volume Manager mirroring technique has the following basic steps:
1. Install multipathing software: In the example, the server is equipped with Sun-badged
HBAs. Sun HBAs allow use of the Sun multipathing solution (Solaris Fibre Channel (FC)
and Storage Multipathing software). However, the example uses DMP for multipathing.
2. Determine migration source LUN information. This step needs to be done to discover
which VxVM object is associated with which device special file. Beyond that you need to
know the size of the source LUNs to provide target LUNs with an appropriate size.
3. Prepare the migration target volumes and investigate their VxVM status.
4. Put the target device under VxVM control.
5. Make the destination volume a mirror member and synchronize the mirror. This is that step
where source data is copied to the target devices.
6. Remove source volumes from the mirror. This step needs to be done to get the source
LUNs out of the VxVM structures.

8.5.3 High-level plan for migration using VxVM mirroring
Table 8-4 Example migration procedure using Solaris with Veritas Storage Foundation
Step 1. Determine the source 1. format Determine the size of the EMC
devices. 2. df migration source file systems
3. vxprint controlled by VxVM and the
4. vxdisk list EMC0_0 migration source mounted file
systems.
Also look at the output of vxprint
to prepare to associate objects
with Solaris device files.
Step 2. Prepare the migration 1. vxdisk list Prepare the target LUNs to be
volumes. used as mirror targets.
Preparation includes verifying
that the target LUNs are visible
from the OS.
Label them if not already done.
Step 3. Investigate the VxVM 1. vxdisk list Determining the DS8000 LUN
migration target device status. IBM_DS8x000_0 size.
2. format
Step 4. Take IBM DS8000 vxdiskadm Take the target devices under
LUNs under control of VxVM. control of VxVM, in this case
using the CLI.
Step 5. Establish the mirrors. 1. vxdiskadm Establish the mirrors using

2. vxtask list vxdiskadm from the CLI and
monitor the progress.
Step 6. Remove the source 1. vxplex -g dg0 -o rm dis Remove the source LUNs from
storage from the mirrors. vol0-01 the mirrors.
2. vxplex -g dg1 -o rm dis
vol1-01
3. vxplex -g dg2 -o rm dis
vol2-01
4. vxprint
Step 7. Remove the source 1. vxdg -g dg0 rmdisk dg001 Remove the EMC LUNs from
LUNs from the VxVM disk 2. vxdg -g dg1 rmdisk dg101 the disk group using vxdg -g
groups. 3. vxdg -g dg2 rmdisk dg201 <dgN> rmdisk.
Step 8. Verify that the LUN is 1. vxdisk list Display the migration status.
removed from the disk group. 2. vxdisk print
8.5.4 Detailed steps for migration using VxVM mirroring

򐂰 Determining source LUN information
򐂰 Preparing the migration volumes
򐂰 Investigating the VxVM migration target device status
򐂰 Taking IBM DS8000 LUNs under control of VxVM
򐂰 Establishing the mirrors
򐂰 Removing the source LUNs from the mirrors
򐂰 Removing the source LUNs from the VxVM disk groups

򐂰 Verifying the migration
Determining source LUN information

To determine the size of the DS8000 migration target volumes, determine the size of the
migration source file systems controlled by VxVM using the following steps:
1. Enter the format command to display the LUNs that are currently visible as shown in
Example 8-26.
Twice as many LUNs show up than expected because DMP is a non-path suppressing
software. This is different from Solaris Fibre Channel (FC) and Storage Multipathing
software. Original devices are still seen even after installation of VxVM and DMP.
Therefore, you see twice the number of disks (once through controller “c2” and once
through controller “c4”) than are assigned to the operating system.
Determine which of these disks represent the migration source volumes and, most
importantly, their size.
Tip: There are two paths representing the EMC LUNs (c2t500604843E0C4BCBd0 and
c4t500604843E0C4BD4d0; size: 6.56 MB) that stay unformatted. These paths are the
EMC Volume Configuration Management Database LUNs (VCMDB) that store host
attachment-specific information. These paths must not be used to store data.
Example 8-26 Display LUNs

# format


2. c2t500604843E0C4BCBd0 <EMC-SYMMETRIX-5568 cyl 14 alt 2 hd 15 sec 64>
/pci@8,600000/fibre-channel@1/fp@0,0/ssd@w500604843e0c4bcb,0
/pci@8,600000/fibre-channel@1/fp@0,0/ssd@w500604843e0c4bcb,1d
9. c4t500604843E0C4BD4d0 <EMC-SYMMETRIX-5568 cyl 14 alt 2 hd 15 sec 64>
/pci@8,700000/fibre-channel@1/fp@0,0/ssd@w500604843e0c4bd4,0

/pci@8,700000/fibre-channel@1/fp@0,0/ssd@w500604843e0c4bd4,1b
2. Use the df command, as shown in Example 8-27, to determine the migration source
mounted file systems. They are already under VxVM control.
Example 8-27 Using the df command to display the currently mounted file systems
# df -k /emc*
/dev/vx/dsk/dg1/vol1 8802304 5095101 3475604 60% /emc1
3. Run the vxprint command to determine the relationship between the VxVM objects and
associate them with the Solaris device files (Example 8-28). In the example, on disk group
dg2, there is one disk/LUN called dg201 with an enclosure-based name of EMC0_2. There is
also one logical volume called vol2.
Example 8-28 Using the command vxprint to show the properties of the VxVM disk groups
# vxprint
Disk group: dg2
dg dg2 dg2 - - - - - -
dm dg201 EMC0_2 - 70624768 - - - -

sd dg201-01 vol2-01 ENABLED 70623232 0 - - -
Disk group: dg0

dg dg0 dg0 - - - - - -
dm dg001 EMC0_0 - 17605888 - - - -

sd dg001-01 vol0-01 ENABLED 17604608 0 - - -
Disk group: dg1

dg dg1 dg1 - - - - - -
dm dg101 EMC0_1 - 17605888 - - - -

sd dg101-01 vol1-01 ENABLED 17604608 0 - - -

4. Use the enclosure-based name as input when running the vxdisk command, which
provides the association to the Solaris device name as seen in Example 8-29. The device
files are at the bottom of the output under the numpaths heading.
Example 8-29 Using the command vxdisk list to show the native device names
# vxdisk list EMC0_0
Device: EMC0_0
devicetag: EMC0_0
type: auto
hostid: SunFire280Rtic2
disk: name=dg001 id=1174034174.32.SunFire280Rtic2
group: name=dg0 id=1174034176.34.SunFire280Rtic2
info: format=cdsdisk,privoffset=256,pubslice=2,privslice=2
flags: online ready private autoconfig autoimport imported
pubpaths: block=/dev/vx/dmp/EMC0_0s2 char=/dev/vx/rdmp/EMC0_0s2
guid: {66fbe82e-1dd2-11b2-9096-0003ba17ecd4}
udid: EMC%5FSYMMETRIX%5F700975%5F750CD000
site: -
version: 3.1
iosize: min=512 (bytes) max=2048 (blocks)
public: slice=2 offset=65792 len=17605888 disk_offset=0
private: slice=2 offset=256 len=65536 disk_offset=0
update: time=1174047713 seqno=0.10
ssb: actual_seqno=0.0
headers: 0 240
configs: count=1 len=48144
logs: count=1 len=7296
Defined regions:
config priv 000048-000239[000192]: copy=01 offset=000000 enabled
config priv 000256-048207[047952]: copy=01 offset=000192 enabled
log priv 048208-055503[007296]: copy=01 offset=000000 enabled
lockrgn priv 055504-055647[000144]: part=00 offset=000000
Multipathing information:
numpaths: 2
c2t500604843E0C4BCBd29s2 state=enabled
c4t500604843E0C4BD4d27s2 state=enabled
The relationships between the Solaris device files and the VxVM disk group and disk
name in the example environment is summarized in Table 8-5.
Table 8-5 Summarizing the findings regarding the source LUNs so far
Native device names Enclosure- Diskgroup Diskname Volname
based name
c2t500604843E0C4BCBd29 EMC0_0 dg0 dg001 vol0

c4t500604843E0C4BD4d27

c4t500604843E0C4BD4d18

c4t500604843E0C4BD4d7
In the example, disk numbers 3 (or 10), 4 (or 11), and 5 (or 12) will be migrated.
5. Select the disk in the menu.

6. Issue the partition to print the partition table.
7. Issue the print to print the partition information.
In the example, note that slice 2 of disk 3 is 33.71 GB. Similarly, disk 4 and 5 are 8.43 GB
each. A slice in Solaris is a partition on the disk. Typically slice 2 refers to the entire disk.
Table 8-6 provides an overview of the source LUNs sizes.
Table 8-6 Showing the source volumes
Native device names Source (enclosure- Size of source (GB)
based naming
scheme)
c2t500604843E0C4BCBd29 EMC0_0 8.43

c4t500604843E0C4BD4d27
c2t500604843E0C4BCBd20 EMC0_1 8.43

c4t500604843E0C4BD4d18
c2t500604843E0C4BCBd9 EMC0_2 33.71

c4t500604843E0C4BD4d7
Preparing the migration volumes

Prepare the target LUNs to be used as mirror targets by performing the following steps:
򐂰 Verify that the target LUNs are visible from the operating system.
򐂰 Label them if you have not already done so,
Investigating the VxVM migration target device status

In this example, three DS8000 migration destination LUNs were requested from the DS8000
storage administrator. These LUNs are slightly larger than the source LUNs: Two 8.43 GB
LUN and one 33.72 GB LUN. Although the two smaller LUNs seem to have the same size as
the source LUNs (8.43 GB), they differ slightly in the number of blocks. Allocation tasks and
complementary tasks such as zoning have been completed successfully.
To investigate the target device status, perform the following steps:

1. Run the vxdisk list command to see an overview of what is currently defined to VxVM
(Example 8-30). The DS8000 LUNs are currently seen by VxVM but not VxVM initialized
yet (online invalid). There are gaps in the naming sequence (IBM_DS8x000_0,
IBM_DS8x000_4 and IBM_DS8x000_5) because more DS8000 LUNs were mapped to
this Sun server, but were later removed.
Example 8-30 Use vxdisk list to see what is currently under control of VxVM
# vxdisk list
EMC0_0 auto:cdsdisk dg001 dg0 online
#

2. Use the vxdisk list command to understand the relationship between the DS8000
enclosure-based names and the underlying Solaris device names as shown in
Example 8-31.
Example 8-31 Using the command vxdisk list to show the native device names
# vxdisk list IBM_DS8x000_0
Device: IBM_DS8x000_0
devicetag: IBM_DS8x000_0
type: auto
info: format=none
flags: online ready private autoconfig invalid
pubpaths: block=/dev/vx/dmp/IBM_DS8x000_0s2 char=/dev/vx/rdmp/IBM_DS8x000_0s2
guid: -
udid: IBM%5F2107%5F75L4741%5F6005076305FFC7860000000000003005
site: -
numpaths: 2
c2t50050763050C0786d5s2 state=enabled
#
type: auto
info: format=none
guid: -
udid: IBM%5F2107%5F75L4741%5F6005076305FFC7860000000000003004
site: -
numpaths: 2
#
type: auto
info: format=none
guid: -
udid: IBM%5F2107%5F75L4741%5F6005076305FFC7860000000000003006
site: -
numpaths: 2

3. Run the format command to discover the size of the target LUNs. Table 8-7 summarizes
the source and destination LUNs in the example. It shows the native device special file
names, and the associated enclosure-based naming scheme names and the respective
size.
Table 8-7 Showing the source and target volumes

Source Size of source (GB) Target Size of target (GB)
EMC0_0 8.43 IBM_DS8x000_0 8.43

c2t500604843E0C4B c2t50050763050C078
CBd29 6d5
c4t500604843E0C4B c4t50050763050C878
D4d27 6d5
EMC0_1 8.43 IBM_DS8x000_5 8.43

c2t500604843E0C4B c2t50050763050C078
CBd20 6d6
c4t500604843E0C4B c4t50050763050C878
D4d18 6d6
EMC0_2 33.71 IBM_DS8x000_4 33.72

c2t500604843E0C4B c2t50050763050C078
CBd9 6d4
c4t500604843E0C4B c4t50050763050C878
D4d7 6d4
Taking IBM DS8000 LUNs under control of VxVM

Place the target devices under VxVM control. There is more than one way to accomplish this
task. This example uses the vxdiskadm CLI command. During this process, you also assign
the DS8000 LUNs that are about to become VxVM-initialized to the existing disk groups. The
following example shows the procedural tasks required to assign enclosure IBM_DS8x000_0
to a VxVM disk group and give it a VxVM name. Although not shown, the same tasks are
successfully completed on enclosures IBM_DS8x000_4 and IBM_DS8x000_5.
To bring the LUNs under VxVM control, perform these steps:

1. Use the vxdiskadm command to initialize the LUN in VxVM (Example 8-32).
Example 8-32 Using the vxdiskadm CLI command to VxVM-initialize IBM_DS8x000_0

Volume Manager Support Operations
Menu: VolumeManager/Disk
1 Add or initialize one or more disks

2 Encapsulate one or more disks
3 Remove a disk
4 Remove a disk for replacement
5 Replace a failed or removed disk
6 Mirror volumes on a disk
7 Move volumes from a disk
8 Enable access to (import) a disk group
9 Remove access to (deport) a disk group
10 Enable (online) a disk device
11 Disable (offline) a disk device
12 Mark a disk as a spare for a disk group
13 Turn off the spare flag on a disk
14 Unrelocate subdisks back to a disk
15 Exclude a disk from hot-relocation use

16 Make a disk available for hot-relocation use
17 Prevent multipathing/Suppress devices from VxVM's view
18 Allow multipathing/Unsuppress devices from VxVM's view
19 List currently suppressed/non-multipathed devices
20 Change the disk naming scheme
21 Get the newly connected/zoned disks in VxVM view
22 Change/Display the default disk layouts
23 Mark a disk as allocator-reserved for a disk group
24 Turn off the allocator-reserved flag on a disk
list List disk information
? Display help about menu

?? Display help about the menuing system
q Exit from menus
Select an operation to perform: 1
2. Select option 1 as shown in Example 8-33.
Example 8-33 Having selected option 1, add or initialize one or more disks
Add or initialize disks
Menu: VolumeManager/Disk/AddDisks
Use this operation to add one or more disks to a disk group. You can
add the selected disks to an existing disk group or to a new disk group
that will be created as a part of the operation. The selected disks may
also be added to a disk group as spares. Or they may be added as
nohotuses to be excluded from hot-relocation use. The selected
disks may also be initialized without adding them to a disk group
leaving the disks available for use as replacement disks.
More than one disk or pattern may be entered at the prompt. Here are
some disk selection examples:
all: all disks

c3 c4t2: all disks on both controller 3 and controller 4, target 2
c3t4d2: a single disk (in the c#t#d# naming scheme)
xyz_0 : a single disk (in the enclosure based naming scheme)
xyz_ : all disks on the enclosure whose name is xyz
Select disk devices to add: [<pattern-list>,all,list,q,?] IBM_DS8x000_0

Here is the disk selected. Output format: [Device_Name]
IBM_DS8x000_0
Continue operation? [y,n,q,?] (default: y) y

You can choose to add this disk to an existing disk group, a
new disk group, or leave the disk available for use by future
add or replacement operations. To create a new disk group,
select a disk group name that does not yet exist. To leave
the disk available for future use, specify a disk group name
of "none".
Which disk group [<group>,none,list,q,?] (default: dg0)

Use a default disk name for the disk? [y,n,q,?] (default: y)
Add disk as a spare disk for dg0? [y,n,q,?] (default: n)
Exclude disk from hot-relocation use? [y,n,q,?] (default: n)
3. Enter the disk names to be initialized as shown in Example 8-34.
Example 8-34 Having selected option 1, add or initialize one or more disks (cont’d)
Add site tag to disk? [y,n,q,?] (default: n)
The selected disks will be added to the disk group dg0 with
default disk names.
IBM_DS8x000_0
Continue with operation? [y,n,q,?] (default: y)

The following disk device has a valid VTOC, but does not appear to have
been initialized for the Volume Manager. If there is data on the disk
that should NOT be destroyed you should encapsulate the existing disk
partitions as volumes instead of adding the disk as a new disk.
Output format: [Device_Name]
IBM_DS8x000_0
Encapsulate this device? [y,n,q,?] (default: y) n
IBM_DS8x000_0
Instead of encapsulating, initialize? [y,n,q,?] (default: n) y

Initializing device IBM_DS8x000_0.
Enter desired private region length

[<privlen>,q,?] (default: 65536)
VxVM NOTICE V-5-2-88
Adding disk device IBM_DS8x000_0 to disk group dg0 with disk
name dg002.
Add or initialize other disks? [y,n,q,?] (default: n)
4. Use the list option on the vxdiskadm CLI command verify that the DS8000 migration
target volumes are under VxVM control as shown in Example 8-35.
Example 8-35 List all devices after they are VxVM-initialized

List disk information
Menu: VolumeManager/Disk/ListDisk
Use this menu operation to display a list of disks. You can
also choose to list detailed information about the disk at
a specific disk device address.
Enter disk device or "all" [<address>,all,q,?] (default: all)
DEVICE DISK GROUP STATUS

EMC0_0 dg001 dg0 online
IBM_DS8x000_0 dg002 dg0 online

c1t0d0 - - online invalid
c1t1d0 - - online invalid
Device to list in detail [<address>,none,q,?] (default: none)
You can also check using the vxdisk list and vxprint commands as shown in
Example 8-36.
Example 8-36 VxVM verification commands

# vxdisk list
IBM_DS8x000_0 auto:cdsdisk dg002 dg0 online
#
## vxprint
Disk group: dg2

dg dg2 dg2 - - - - - -
dm dg201 EMC0_2 - 70624768 - - - -

dm dg202 IBM_DS8x000_4 - 70647552 - - - -

sd dg201-01 vol2-01 ENABLED 70623232 0 - - -
Disk group: dg0

dg dg0 dg0 - - - - - -
dm dg001 EMC0_0 - 17605888 - - - -

dm dg002 IBM_DS8x000_0 - 17607808 - - - -

sd dg001-01 vol0-01 ENABLED 17604608 0 - - -
Disk group: dg1

dg dg1 dg1 - - - - - -
dm dg101 EMC0_1 - 17605888 - - - -

dm dg102 IBM_DS8x000_5 - 17607808 - - - -

sd dg101-01 vol1-01 ENABLED 17604608 0 - - -
Establishing the mirrors

To establish the VxVM mirrors, perform the following steps:
1. Run the vxdiskadm CLI command and select option 6, Mirror volumes on a disk as
shown in Example 8-37. This example refers to dg0 and vol0.
Example 8-37 Establishing mirroring

Select an operation to perform: 6
Mirror volumes on a disk

Menu: VolumeManager/Disk/Mirror
This operation can be used to mirror volumes on a disk. These
volumes can be be mirrored onto another disk or onto any
available disk space. Volumes will not be mirrored if they are
already mirrored. Also, volumes that are comprised of more than
one subdisk will not be mirrored.
Mirroring volumes from the boot disk will produce a disk that
can be used as an alternate boot disk.
At the prompt below, supply the name of the disk containing the
volumes to be mirrored.
Enter disk name [<disk>,list,q,?] list

Disk group: dg2
DM NAME DEVICE TYPE PRIVLEN PUBLEN STATE
dm dg201 EMC0_2 auto 65536 70624768 -

dm dg202 IBM_DS8x000_4 auto 65536 70647552 -
Disk group: dg0
dm dg001 EMC0_0 auto 65536 17605888 -

dm dg002 IBM_DS8x000_0 auto 65536 17607808 -
Disk group: dg1
dm dg101 EMC0_1 auto 65536 17605888 -

dm dg102 IBM_DS8x000_5 auto 65536 17607808 -
Enter disk name [<disk>,list,q,?] dg001

You can choose to mirror volumes from disk dg001 onto any
available disk space, or you can choose to mirror onto a specific
disk. To mirror to a specific disk, select the name of that disk.
To mirror to any available disk space, select "any".
Enter destination disk [<disk>,list,q,?] (default: any) dg002

The requested operation is to mirror all volumes on disk dg001
in disk group dg0 onto available disk space on disk dg002.
VxVM NOTICE V-5-2-3650 This operation can take a long time to complete.
Continue with operation? [y,n,q,?] (default: y)

VxVM vxmirror INFO V-5-2-22 Mirror volume vol0 ...
2. Repeat step 1 for all other disks.

3. Use the vxtask list command to monitor the progress of the mirroring activity
(Example 8-38).
Example 8-38 Using vxtask list to display the mirroring progress

# vxtask list
TASKID PTID TYPE/STATE PCT PROGRESS
170 ATCOPY/R 13.58% 0/17604608/2390016 PLXATT vol1 vol1-02 dg1
#
4. The mirrors are in sync when the vxtask list command no longer shows any ongoing
activity (Example 8-39).
5. Enter vxprint and make sure that new plexes and subdisks have been created in each
disk group.
Example 8-39 Using vxtask list and then vxprint

# vxtask list
TASKID PTID TYPE/STATE PCT PROGRESS
# vxprint
Disk group: dg2

dg dg2 dg2 - - - - - -
dm dg201 EMC0_2 - 70624768 - - - -

dm dg202 IBM_DS8x000_4 - 70647552 - - - -

sd dg201-01 vol2-01 ENABLED 70623232 0 - - -
sd dg202-01 vol2-02 ENABLED 70623232 0 - - -
Disk group: dg0

dg dg0 dg0 - - - - - -
dm dg001 EMC0_0 - 17605888 - - - -

dm dg002 IBM_DS8x000_0 - 17607808 - - - -

sd dg001-01 vol0-01 ENABLED 17604608 0 - - -
sd dg002-01 vol0-02 ENABLED 17604608 0 - - -

Disk group: dg1

dg dg1 dg1 - - - - - -
dm dg101 EMC0_1 - 17605888 - - - -

dm dg102 IBM_DS8x000_5 - 17607808 - - - -

sd dg101-01 vol1-01 ENABLED 17604608 0 - - -
sd dg102-01 vol1-02 ENABLED 17604608 0 - - -
Removing the source LUNs from the mirrors

Remove the source LUNs from the mirrors using the vxplex command (Example 8-40). In the
example, a subsequent vxprint shows the results of the vxplex operation against dg0. The
source EMC LUN has been removed from the mirror but is still part of the disk group.
Example 8-40 Using vxplex to get the EMC LUN out of the mirror
# vxplex -g dg0 -o rm dis vol0-01
#
# vxprint -g dg0
dg dg0 dg0 - - - - - -
dm dg001 EMC0_0 - 17605888 - - - -

dm dg002 IBM_DS8x000_0 - 17607808 - - - -

sd dg002-01 vol0-02 ENABLED 17604608 0 - - -
The same task needs to be done for dg1 and dg2, as shown in Example 8-41.
Example 8-41 Using vxplex to get the EMC LUN out of the mirrors
# vxprint
Disk group: dg2

dg dg2 dg2 - - - - - -
dm dg201 EMC0_2 - 70624768 - - - -

dm dg202 IBM_DS8x000_4 - 70647552 - - - -

sd dg202-01 vol2-02 ENABLED 70623232 0 - - -
Disk group: dg0

dg dg0 dg0 - - - - - -
dm dg001 EMC0_0 - 17605888 - - - -

dm dg002 IBM_DS8x000_0 - 17607808 - - - -

sd dg002-01 vol0-02 ENABLED 17604608 0 - - -
Disk group: dg1

dg dg1 dg1 - - - - - -
dm dg101 EMC0_1 - 17605888 - - - -

dm dg102 IBM_DS8x000_5 - 17607808 - - - -

sd dg102-01 vol1-02 ENABLED 17604608 0 - - -
Removing the source LUNs from the VxVM disk groups

Remove the source EMC LUNs from the disk group by entering vxdg -g <dgN> rmdisk ... , 0
<= N <= 2 as seen in Example 8-42. After the process is complete, display the disk groups
again to verify that the EMC LUNs are removed.
Example 8-42 Removing EMC LUNs from the disk groups and displaying the results
# vxdg -g dg0 rmdisk dg001
# vxdisk list
EMC0_0 auto:cdsdisk - - online
Verifying the migration

Verify that the migration is complete by entering vxprint (Example 8-43).
Example 8-43 Entering vxprint to see that EMC LUNs are not contained in any of the disk groups
# vxprint
Disk group: dg2

dg dg2 dg2 - - - - - -
dm dg202 IBM_DS8x000_4 - 70647552 - - - -

sd dg202-01 vol2-02 ENABLED 70623232 0 - - -
Disk group: dg0

dg dg0 dg0 - - - - - -
dm dg002 IBM_DS8x000_0 - 17607808 - - - -

sd dg002-01 vol0-02 ENABLED 17604608 0 - - -
Disk group: dg1

dg dg1 dg1 - - - - - -
dm dg102 IBM_DS8x000_5 - 17607808 - - - -

sd dg102-01 vol1-02 ENABLED 17604608 0 - - -
The migration is now complete and the data is located exclusively on DS8000 LUNs. There
might be some additional tasks left beyond the data migration itself, such as disconnecting
the EMC LUNs from the system.
8.6 Data migration using HP-UX Volume Manager mirroring

This section addresses the tasks required to migrate existing EMC LUNs to new DS8000
LUNs, using the HP-UX Volume Manager mirroring capabilities. The process addressed here
is fairly generic. The number of LUNs and the type of storage subsystem they originate from
do not really matter: The process is the same for one or multiple LUNs from any subsystem.
Depending on the specifics of your environment, the process can be disruptive to
applications, requiring reboots and file system unmounts.
In the scenario that follows, the volume to be migrated is part of a volume group (VG). In this
example, the EMC storage LUN to be migrated is integrated into a VG called vg_emc_to_ibm.

8.6.1 High-level plan for migration using HP_UX Volume Manager mirroring
Table 8-8 Example migration procedure using HP-UX Volume Manager
Step 1. Get an overview of the ioscan -fnC disk Discover the DS8000 LUNs on
attached disks. the HP-UX host system.
Step 2. Create the device insf -e You must create the device
special files. special files before you can
work with the DS8000 LUNs.
Step 3. Verify that the device ioscan -fnC disk Rescan.

special files were created.
Step 4. Prepare the target 1. pvcreate Use one LUN as an argument

volumes to be included into a /dev/rdsk/c125t0d0 for pvcreate to make the new
volume group. 2. vgdisplay -v device ready to be used in a
/dev/vg_emc_to_ibm volume group (VG).
Step 5. Integrate the target vgextend /dev/vg_emc_to_ibm Bring an IBM storage LUN into
volumes into the respective /dev/dsk/c126t0d0 VG vg_emc_to_ibm.
volume group.
Step 6. Assign an alternative vgextend /dev/vg_emc_to_ibm For redundancy, provide an

path for the LUN that just /dev/dsk/c125t0d0 alternative path for the DS8000
became integrated. LUN.
Step 7. Set up a mirror. lvextend -m 1 Create a mirror for the logical

/dev/vg_emc_to_ibm/lvol1 volume and force that mirror to
/dev/dsk/c125t0d0 be on /dev/dsk/c125t0d0 (one
of the paths to the IBM LUN).
Step 8. Remove the source 1. vgreduce Remove the alternative link to

LUN. /dev/vg_emc_to_ibm the EMC device.
/dev/dsk/c123t1d4 Reduce the number of mirror
2. lvreduce -m 0 copies.
/dev/vg_emc_to_ibm/lvol1 Remove the last remaining path
/dev/dsk/c121t1d6 to the EMC LUN.
3. vgreduce
/dev/vg_emc_to_ibm
/dev/dsk/c121t1d6
Step 9. Verify that the mirror is vgdisplay -v Verify that no EMC device
removed. /dev/vg_emc_to_ibm special file is being referred to.
8.6.2 Detailed steps for migration using HP-UX Volume Manager mirroring
򐂰 Getting an overview of the attached disks
򐂰 Creating device special files
򐂰 Preparing the target volumes for inclusion in a volume group
򐂰 Integrating the target volumes into the respective volume group
򐂰 Assigning an alternative path to the LUN
򐂰 Setting up a mirror
򐂰 Removing the source LUN

Getting an overview of the attached disks
Check which disks in your environment are visible by the HP-UX host using the ioscan -fnC
disk command. In the example environment shown in Example 8-44, apart from the EMC
LUNs, the HP-UX host sees four DS8000 LUNs. You see twice the number of LUNs than are
available because the storage administrator assigned two LUNs that are presented through
two ESS ports.
Example 8-44 Using ioscan to show the LUNs that are visible to the OS
# ioscan -fnC disk
Class I H/W Path Driver S/W State H/W Type Description
==========================================================================
disk 0 0/0/1/1.2.0 sdisk CLAIMED DEVICE FUJITSU MAJ3182MC
/dev/dsk/c1t2d0 /dev/rdsk/c1t2d0
disk 1 0/0/2/1.2.0 sdisk CLAIMED DEVICE HP DVD-ROM 305
disk 266 0/4/0/0.18.50.0.36.0.0 sdisk CLAIMED DEVICE IBM 2107900
disk 262 0/4/0/0.18.60.0.0.0.0 sdisk CLAIMED DEVICE EMC SYMMETRIX
Creating device special files

You must create the device special files for the DS8000 LUNs before you can work with those
LUNs. To create the files, perform the following steps:
1. Run the insf -e command as shown in Example 8-45.
Example 8-45 Using insf -e to create the device special files

# insf -e
insf: Installing special files for btlan instance 0 address 0/0/0/0
insf: Installing special files for sctl instance 0 address 0/0/1/0.7.0
insf: Installing special files for sdisk instance 0 address 0/0/1/1.2.0
insf: Installing special files for sdisk instance 1 address 0/0/2/1.2.0
insf: Installing special files for asio0 instance 0 address 0/0/4/1
insf: Installing special files for hub instance 0 address 0/2/0/0.1
insf: Installing special files for btlan instance 1 address 0/3/0/0/4/0
insf: Installing special files for sdisk instance 266 address 0/4/0/0.18.50.0.36
.0.0
insf: Installing special files for sctl instance 23 address 0/4/0/0.18.50.255.0.
0.0
.0.0

0.0
insf: Installing special files for sdisk instance 262 address 0/4/0/0.18.60.0.0.
0.0
1.4
0.0
.0.0
0.0
.0.0
0.0
0.0
1.6
0.0
insf: Installing special files for pseudo driver cn
insf: Installing special files for pseudo driver mm
insf: Installing special files for pseudo driver devkrs
insf: Installing special files for pseudo driver ptym
insf: Installing special files for pseudo driver ptys
insf: Installing special files for pseudo driver ip
insf: Installing special files for pseudo driver arp
insf: Installing special files for pseudo driver rawip
insf: Installing special files for pseudo driver tcp
insf: Installing special files for pseudo driver udp
insf: Installing special files for pseudo driver stcpmap
insf: Installing special files for pseudo driver nuls
insf: Installing special files for pseudo driver netqa
insf: Installing special files for pseudo driver dmem
insf: Installing special files for pseudo driver diag0
insf: Installing special files for pseudo driver telm
insf: Installing special files for pseudo driver tels
insf: Installing special files for pseudo driver tlclts
insf: Installing special files for pseudo driver tlcots
insf: Installing special files for pseudo driver iomem
insf: Installing special files for pseudo driver tlcotsod
insf: Installing special files for pseudo driver dmp
insf: Installing special files for pseudo driver vols
insf: Installing special files for pseudo driver dev_config
insf: Installing special files for pseudo driver strlog
insf: Installing special files for pseudo driver sad
insf: Installing special files for pseudo driver echo
insf: Installing special files for pseudo driver dlpi
insf: Installing special files for pseudo driver ptm
insf: Installing special files for pseudo driver pts
insf: Installing special files for pseudo driver beep
insf: Installing special files for pseudo driver framebuf
insf: Installing special files for pseudo driver klog

insf: Installing special files for pseudo driver sy
insf: Installing special files for pseudo driver kepd
insf: Installing special files for pseudo driver root
2. Enter the ioscan -fnC disk command again to verify that the device special files were
created (Example 8-46).
Example 8-46 Using ioscan -fnC disk

# ioscan -fnC disk
Class I H/W Path Driver S/W State H/W Type Description
==========================================================================
disk 0 0/0/1/1.2.0 sdisk CLAIMED DEVICE FUJITSU MAJ3182MC
disk 1 0/0/2/1.2.0 sdisk CLAIMED DEVICE HP DVD-ROM 305
Preparing the target volumes for inclusion in a volume group

After the device special files are available, use one of them as argument for the pvcreate
command as shown in Example 8-47. This command makes the new device ready to be used
in a volume group (VG).
Example 8-47 Using pvcreate to make a LUN ready to be integrated into a volume group
# pvcreate /dev/rdsk/c125t0d0
Physical volume "/dev/rdsk/c126t0d0" has been successfully
created
Issue the command vgdisplay -v /dev/vg_emc_to_ibm to see the following characteristics of

this particular VG:
򐂰 Size
򐂰 Physical volumes (PVs)
򐂰 Physical elements
򐂰 Logical volumes (LVs).
Example 8-48 shows the output of the vgdisplay -v /dev/vg_emc_to_ibm command.
Example 8-48 Issuing vgdisplay -v /dev/vg_emc_to_ibm

# vgdisplay -v /dev/vg_emc_to_ibm
--- Volume groups ---

VG Name /dev/vg_emc_to_ibm
VG Write Access read/write
VG Status available
Max LV 255
Cur LV 1
Open LV 1
Max PV 16
Cur PV 1
Act PV 1
Max PE per PV 2157
VGDA 2
PE Size (Mbytes) 4
Total PE 2156
Alloc PE 2156
Free PE 0
Total PVG 0
Total Spare PVs 0
Total Spare PVs in use 0
--- Logical volumes ---

LV Name /dev/vg_emc_to_ibm/lvol1
LV Status available/syncd
LV Size (Mbytes) 8624
Current LE 2156
Allocated PE 2156
Used PV 1
--- Physical volumes ---

PV Name /dev/dsk/c121t1d6
PV Name /dev/dsk/c123t1d4 Alternate Link
PV Status available
Total PE 2156
Free PE 0
Autoswitch On
Integrating the target volumes into the respective volume group

Use the vgextend command to bring a DS8000 target LUN into the VG vg_emc_to_ibm as
Example 8-49 Using the command vgextend to add an IBM LUN into the existing VG
# vgextend /dev/vg_emc_to_ibm /dev/dsk/c126t0d0
vgextend: Warning: Max_PE_per_PV for the volume group (2157) too small for this PV
(2303).
Using only 2157 PEs from this physical volume.
Current path "/dev/dsk/c123t1d4" is an alternate link, skip.
Volume group "/dev/vg_emc_to_ibm" has been successfully extended.
Volume Group configuration for /dev/vg_emc_to_ibm has been saved in
/etc/lvmconf/vg_emc_to_ibm.conf
# vgdisplay
VG Name /dev/vg00
VG Status available

Max LV 255
Cur LV 11
Open LV 11
Max PV 16
Cur PV 1
Act PV 1
Max PE per PV 4350
VGDA 2
PE Size (Mbytes) 4
Total PE 4340
Alloc PE 2788
Free PE 1552
Total PVG 0
Total Spare PVs 0
VG Status available
Max LV 255
Cur LV 1
Open LV 1
Max PV 16
Cur PV 2
Act PV 2
Max PE per PV 2157
VGDA 4
PE Size (Mbytes) 4
Total PE 4313
Alloc PE 2156
Free PE 2157
Total PVG 0
Total Spare PVs 0
The warning message “vgextend: Warning: Max_PE_per_PV for the volume group (2157)
too small for this PV (2303)“ is displayed because the DS8000 LUN size exceeds the size
of the EMC LUN. When creating a volume group, the number of physical elements is set to a
default value. If the number of physical elements calculated when creating the volume group
exceeds that default value, the calculated number is used.
To avoid this problem, perform the following steps:

򐂰 Make sure that the administrator defines enough physical elements. Check the elements
using option -e when running the vgcreate command.
򐂰 Create the target LUNs so that their number of physical elements do not exceed the
number of physical elements that is valid for this volume group. In this case, the target
source LUNs would need to the same size as the source LUNs.
You can also accept that some space on the target volume is wasted.
Tip: More recent versions of HP-UX (HP-UX 11i v3 and later) allow you to modify the
physical element number dynamically by issuing the vgmodify command.

Assigning an alternative path to the LUN
For redundancy, provide and verify an alternative path for the DS8000 LUN as shown in
Example 8-50.
Example 8-50 Provide an alternative path for the target volume and verify successful completion
# vgextend /dev/vg_emc_to_ibm /dev/dsk/c125t0d0
vgextend: Warning: Max_PE_per_PV for the volume group (2157) too small for this
PV (2303).
Using only 2157 PEs from this physical volume.
Current path "/dev/dsk/c123t1d4" is an alternate link, skip.
Volume group "/dev/vg_emc_to_ibm" has been successfully extended.
Volume Group configuration for /dev/vg_emc_to_ibm has been saved in /etc/lvmconf
/vg_emc_to_ibm.conf
VG Status available
Max LV 255
Cur LV 1
Open LV 1
Max PV 16
Cur PV 2
Act PV 2
Max PE per PV 2157
VGDA 4
PE Size (Mbytes) 4
Total PE 4313
Alloc PE 2156
Free PE 2157
Total PVG 0
Total Spare PVs 0

Current LE 2156
Allocated PE 2156
Used PV 1

PV Status available
Total PE 2156
Free PE 0
Autoswitch On
PV Status available
Total PE 2157
Free PE 2157
Autoswitch On

In the example, /dev/dsk/c125t0d0 is labeled Alternate Link. It is labeled this way because
/dev/dsk/c126t0d0 and /dev/dsk/c125t0d0 are two paths to the same volume rather than
two separate volumes.
Setting up a mirror
Use the lvextend -m 1 command to create a mirror for the logical volume. Force that mirror to
be on /dev/dsk/c125t0d0 (one of the paths to the IBM LUN) as shown in Example 8-51. Wait
until the mirror becomes established.
Example 8-51 Mirror to the target device

# lvextend -m 1 /dev/vg_emc_to_ibm/lvol1 /dev/dsk/c125t0d0
Device file path "/dev/dsk/c125t0d0" is an alternate path
to the Physical Volume. Using Primary Link "/dev/dsk/c126t0d0".
The newly allocated mirrors are now being synchronized. This operation will
take some time. Please wait ....
Logical volume "/dev/vg_emc_to_ibm/lvol1" has been successfully extended.
/vg_emc_to_ibm.conf
Removing the source LUN

The following steps address how to remove the source (EMC) LUN:
1. Remove the alternate link to the EMC device using the vgreduce command as shown in
Example 8-52.
Example 8-52 Using the command vgreduce to remove one path to the EMC LUN
# vgreduce /dev/vg_emc_to_ibm /dev/dsk/c123t1d4
Device file path "/dev/dsk/c123t1d4" is an alternate path.
Volume group "/dev/vg_emc_to_ibm" has been successfully reduced.
/vg_emc_to_ibm.conf
2. Use the lvreduce command to reduce the number of mirror copies so that the EMC LUN
is not part of the mirror (Example 8-53).
Example 8-53 Using the command lvreduce to remove the EMC LUN from the mirror
#lvreduce -m 0 /dev/vg_emc_to_ibm/lvol1 /dev/dsk/c121t1d6
Logical volume "/dev/vg_emc_to_ibm/lvol1" has been successfully reduced.
/vg_emc_to_ibm.conf
3. Remove the last remaining path to the EMC LUN with the vgreduce command as shown in
Example 8-54.
Example 8-54 Remove the last remaining path to the EMC LUN
vgreduce /dev/vg_emc_to_ibm /dev/dsk/c121t1d6
Volume group "/dev/vg_emc_to_ibm" has been successfully reduced.
/vg_emc_to_ibm.conf
#

Verify that no EMC device special file is displayed by issuing the vgdisplay command
(Example 8-55).
Example 8-55 Using vgdisplay to display VG components

VG Status available
Max LV 255
Cur LV 1
Open LV 1
Max PV 16
Cur PV 1
Act PV 1
Max PE per PV 2157
VGDA 2
PE Size (Mbytes) 4
Total PE 2157
Alloc PE 2156
Free PE 1
Total PVG 0
Total Spare PVs 0

Current LE 2156
Allocated PE 2156
Used PV 1

PV Status available
Total PE 2157
Free PE 1
Autoswitch On
8.7 Data migration using AIX LVM mirroring

This section addresses mirroring using the AIX Logical Volume Manager (LVM) to migrate
data from an ESS to a DS8000.
There are two possible methods to migrate using AIX LVM mirroring:
򐂰 Using migratepv -l
򐂰 Using mklvcopy and syncvg

You can connect the replacement DS8000 in parallel with the existing ESS. Assign the LUNs
from the DS8000 to the same compatible host server HBAs that the ESS LUNs are assigned
to. In this way, you can mirror LUNs or volumes on the existing unit to the replacement unit
using software (LVM) techniques.
After the volumes are mirrored, break the mirror from the existing unit and remove the old
unit. Prepare the DS8000 with the correct Array to LUN spread and LUN size considerations
as it applies to the current setup of the ESS. For more information about DS8000 preparation,
see 2.4, “Preparing DS8000 for data migration” on page 27
8.7.1 ESS to DS8000 migration under AIX LVM test environment

The physical test environment shown in Figure 8-1 on page 269 is composed of the following
components:
򐂰 IBM eServer™ pSeries® 615 Model 6C3 (now discontinued), 7029-6C3 AIX level 5.3.0.0,
maintenance level 04
– IBM SDD PCM for AIX V53 version 2.1.2.0 Disk Device driver
– Emulex HBAs running firmware version 9.1.2.14, Feature Code 5704, Part Number
00P4295
򐂰 ESS model 800 Storage running microcode level 2.4.4.45
򐂰 DS8100 model 921 running firmware level SEA 5.2.400.437, bundle 6.2.400.76
򐂰 Brocade Switch type 44, Fabric OS firmware version 5.1.0b
The example environment is three existing ESS data volumes (left side of the diagram in solid
lines) configured on the Windows server and running live applications. Introduce the DS8000
into the environment and migrate the 2105 volumes (source volumes) to the DS8000 volumes
(target volumes). The three DS8000 volumes were created of equal or greater capacity using
type ds as in the DS8000. The target volume capacities (sizes, are shown on the right side of
the diagram in dotted lines.

ESS 800 IBM DS8100
Tota lS to ra g e
T otalStor age
4.736 GB P615 model 6C3 5.12 GB

1 2
1 2
M F G IN T
1
HMC
1
1
HMC OK OK
2
2
1 2 3 4 5 6
4.736 GB 5.12 GB
.896 GB 1.24 GB
Figure 8-24 Migration test environment using the AIX LVM
8.7.2 High-level migration plan using the AIX LVM migratepv -l

The high-level steps for a migration based on the migratepv -l command are shown in
Table 8-9. The table illustrates an ESS to DS8000 migration example. You can apply the
same process to other open system storage platforms.
Table 8-9 Example migration procedure using migratepv -l
Step1. Identify the ESS source lsvpcfg Gather information about the
LUNs. AIX host, such as the number of
ESS LUNs and sizes.
Step 2. Assign the DS8000 chvolgrp -dev image_id On the DS8000, use the
LUNs to the host. -action add-volume vol_# V3 DS8000 CLI (dscli) to assign
the LUNs. The image_id is the
DS8000 ID and vol_# is the
number of the DS8000 volume
in hex.
Step 3. Discover the DS8000 cfgmgr On the AIX host, discover the
LUNs. newly assigned LUNs.
Step 4. Determine the lsvpcfg This command works using the

difference between the ESS SDD driver on the AIX host. The
and DS8000 LUNs. source LUNs are ESS and the
target LUNs are DS8000. Look
at the serial numbers of the
LUNs in the output.

Step 5. Identify the sizes of the bootinfo -s vpath# Where # is the number of the
DS8000 target LUNs. vpath.
Step 6. Move the DS8000 extendvg vg-name vpath# Where vg_name is the name of
LUNs into the appropriate VGs. the VG and # is the number of
the vpath.
Step 7. Verify that the DS8000 lsvg -p vg_name Where vg_name is the name of
LUNs are added to the VG. the VG.
Step 8. Identify the logical lsvg -l vg_name Where vg_name is the name of
volumes (LVs) to migrate. the VG.
Step 9. Copy LV data from the migratepv -l lv_name Where lv_name is the name of
ESS source LUNs to the the LV.
DS8000 target LUNs.
Step 10. Verify that the LUNs lsvg -p vg_name Where vg_name is the name of
are copied. lspv -l vpath# the VG and # is the number of
the vpath.
Step 11. Remove the ESS reducevg vg_name vpath# Where vg_name is the name of
source LUNs from the VGs and lsvg -p vg_name the VG and # is the number of
verify that the source ESS the vpath.
LUNs are removed from the
VG.
Step 12. Delete the device rmdev -dl vpath# Where # is the number of the
definitions from the host ODM. rmdev -dl hdisk# vpath and appropriate hdisk.
Step 13. Verify that the device lsdev -Cc disk Check that there are no defined
definitions are removed disks.
Step 14. Remove the source n/a Removes data path between
zone definition from the switch AIX host and disk subsystem
fabric.
Step 15. In the ESS, unassign 1. Click Modify Volume This sequence is done from the
the LUNs from the host server. Assignments. ESS Specialist GUI.
2. In the Volume assignments
window, click First from the
Host Nicknames list.
3. Click Perform Sort.
4. Look for the first assigned
volume assigned to the
host.
5. Press the Ctrl key and click
the volume.
6. In the Action list, select
Unassign selected
volume(s) to target hosts.
7. In the Target Hosts list,
select all the targeted
volumes using the Ctrl key.
8. Click Perform
Configuration Update.
9. Click OK.

8.7.3 Detailed steps using migratepv -l
The migration procedure has the following detailed steps, using AIX level 5.2 with both the
2105 and the 2107 SDD drivers:
򐂰 Identifying the ESS source LUNs in the targeted VGs
򐂰 Assigning the DS8000 LUNs to the AIX host
򐂰 Discovering the DS8000 target LUNs
򐂰 Determining the differences between the ESS and DS8000 LUNs
򐂰 Identifying the sizes of the DS8000 LUNs
򐂰 Moving the DS8000 LUNs into the VGs appropriately
򐂰 Verifying that the DS8000 LUNs are added to the VG
򐂰 Identifying the LVs to migrate
򐂰 Copying LV data from the source LUNs to the DS8000 target LUNs
򐂰 Verifying that the LUNs are copied
򐂰 Removing the ESS source LUNs from the VGs
򐂰 Deleting the device definitions from the host ODM
򐂰 Verifying that the device definitions are removed
򐂰 Removing the source zone definition
򐂰 Unassigning the source LUNs from the host server
Identifying the ESS source LUNs in the targeted VGs

By issuing lsvg -p against the target vg_name, you can determine the size of the LUNs that
you need to assign on the DS8000. See Example 8-56.
Multiply the number of physical partitions (PPs) for the vpath by the PP size of the volume
group. This calculation gives you the LUN size in MB. The formula is:
(LUNPPs) * vgPPsize = LUN size
If you take the Total PPs for vpath2 and multiply it by the vg PP size of 64, you get 4736 MB:
(74) * (64) = 4736 MB
Example 8-56 Results from running the lsvg -p command

# lsvg -p chuckvg
chuckvg:
PV_NAME PV STATE TOTAL PPs FREE PPs FREE DISTRIBUTION
vpath2 active 74 0 00..00..00..00..00
vpath3 active 74 1 00..00..00..00..01
vpath4 active 14 1 00..00..00..00..01
Assigning the DS8000 LUNs to the AIX host

The example scenario has the volumes and volgrp already created. Assign the new volumes
4004 - 4006 to V3, and initialize the host connections to the DS8000 using the dscli
command as shown in Example 8-57.
Example 8-57 The dscli command to assign storage to the AIX host
mkfbvol -dev IBM.2107-75L4741 -extpool P4 -name 8300_#h -type ds -cap 5 4004-4005
mkfbvol -dev IBM.2107-75L4741 -extpool P4 -name 8300_#h -type ds -cap 1 4006
mkvolgrp -dev IBM.2107-75L4741 IBM.2107-7581981 scsimask -volume 4004-4006 V3
mkhostconnect -dev IBM.2107-75L4741 -wwname 10000000C93E007C -hosttype pSeries
-volgrp V5 p615-tic-6_A0
mkhostconnect -dev IBM.2107-75L4741 -wwname 10000000C93E0059 -hosttype pSeries

To verify that LUNs are assigned to V3, issue the showvolgrp command as shown in
Example 8-58.
Example 8-58 Displaying LUNs in a volgrp using the dscli

IBM.2107-75L4741
Name p615_tic6_vg
ID V3
Type SCSI Mask
Vols 4004 4005 4006
LUNs 4004-4005 are assigned to volgrp V3.
Discovering the DS8000 target LUNs

Assign the DS8000 LUNs and run cfgmgr on the AIX host system to discover the newly
assigned LUNs.
Determining the differences between the ESS and DS8000 LUNs

Identify the DS8000 and ESS LUNs by the LUN ID displayed in the output of the lsvpcfg
Example 8-59 Identifying ESS and DS8000 LUNs

# lsvpcfg
vpath2 (Avail pv chuckvg) 70022665 = hdisk19 (Avail ) hdisk22 (Avail ) hdisk25
(Avail ) hdisk34 (Avail )
vpath5 (Avail ) 75L47414005 = hdisk14 (Avail ) hdisk17 (Avail ) hdisk29 (Avail )
hdisk32 (Avail )
vpath6 (Avail ) 75L47414006 = hdisk15 (Avail ) hdisk18 (Avail ) hdisk30 (Avail )
hdisk33 (Avail )
vpath7 (Avail pv ) 75L47414004 = hdisk16 (Avail ) hdisk28 (Avail ) hdisk37 (Avail )
hdisk38 (Avail
)
The ESS LUN IDs consist of the first three digits, which giving the LUN hex ID, and the last
five digits that are the ESS serial number. The DS8000 LUNs are the opposite way: The
DS8000 serial number is the first eight digits, and the last four digits are the LUN hex ID. For
example, vpath 5 is a DS8000 LUN, so 75L4741 is the serial number of the unit and 4005 is
the LUN ID. In the LUN ID, 40 is the LSS and 05 is the LUN ID in hex.
Identifying the sizes of the DS8000 LUNs

You need to know the sizes of the DS8000 LUNs to move them into the VGs. Remember that
the DS8000 target LUN should be at least the same size or larger than the source ESS LUN
when using the migratepv command. If you choose to use the mklvcopy command, check that
there is enough space in the VG on the DS8000 LUNs collectively to create the mirrors. Issue
the bootinfo -s vpath/hdisk command to identify the sizes in MB (Example 8-60).
Example 8-60 Output showing the results of the bootinfo command

# bootinfo -s vpath5
5120

1024
5120
Moving the DS8000 LUNs into the VGs appropriately

Issue the extendvg command to move the DS8000 LUNs into the targeted volume groups as
Example 8-61 Using the extendvg command to move LUNs into a VG

extendvg chuckvg vpath5 vpath6 vpath7
0516-1254 extendvg:Changing the PVID in the ODM
Verifying that the DS8000 LUNs are added to the VG

To verify that the vpaths are added to the VG correctly, run the lsvg -p command as shown in
Example 8-62.
Example 8-62 Output results from using the lsvg -p command

# lsvg -p chuckvg
chuckvg:
vpath2 active 74 0 00..00..00..00..00
vpath3 active 74 1 00..00..00..00..01
vpath4 active 14 1 00..00..00..00..01
vpath5 active 79 79 16..16..15..16..16
vpath6 active 15 15 03..03..03..03..03
vpath7 active 79 79 16..16..15..16..16
Identifying the LVs to migrate

To identify the LVs to migrate from the source ESS LUNs to the target DS8000 LUNs, issue
the command lsvg - l as shown in Example 8-63.
Example 8-63 The output results by issuing the lsvg -l command against a VG
# lsvg -l chuckvg
chuckvg:
ess5GBlv1 jfs2 73 73 1 open/syncd /ess5GB1
ess1GBlv jfs2 13 13 1 open/syncd /ess1GB

Copying LV data from the source LUNs to the DS8000 target LUNs
Use the migratepv command to accomplish the migration. To move physical partitions from
vpath2 to vpath5, issue the migratepv -l command against the VG as shown in
Example 8-64.
Note: Do not use the migratepv -p command. An error can leave the logical volume with
data on both the target and source LUNs in an unknown state.
Example 8-64 Using the migratepv -l command to migrate data

# migratepv -l ess5GBlv1 vpath2 vpath5
Verifying that the LUNs are copied

When the migration is completed, verify that the data contents on vpath2 are migrated to the
target DS8000 LUN (vpath5). You can check by issuing the lsvg -p command as shown in
Example 8-65.
Example 8-65 Results from issuing the lsvg -p command

# lsvg -p chuckvg
chuckvg:
vpath2 active 74 73 15..15..14..15..14
vpath3 active 74 1 00..00..00..00..01
vpath4 active 14 1 00..00..00..00..01
vpath5 active 79 6 00..00..00..00..06
vpath6 active 15 15 03..03..03..03..03
vpath7 active 79 79 16..16..15..16..16
Notice that vpath2 now has 73 PPs free and vpath5 has the contents on it.
Another way to verify that the data is by issuing the lspv -l command against the source
and target vpaths as shown in Example 8-66.
Example 8-66 Results from issuing the lspv -l command against the vpath
# lspv -l vpath5
vpath5:
LV NAME LPs PPs DISTRIBUTION MOUNT POINT
ess5GBlv1 73 73 16..16..15..16..10 /ess5GB1
# lspv -l vpath2
vpath2:
loglv00 1 1 00..00..00..00..01 N/A
Notice that there is still one PP on vpath2 that needs to be moved.

Move loglv00 from the targeted ESS LUN vpath2 to the targeted DS8000 LUN by following
the process shown in Example 8-67.
Example 8-67 Using the migratepv -l command and the output results
# migratepv -l loglv00 vpath2 vpath5
# lspv -l vpath2
# lspv -l vpath5
vpath5:
ess5GBlv1 73 73 16..16..15..16..10 /ess5GB1
loglv00 1 1 00..00..00..00..01 N/A
Removing the ESS source LUNs from the VGs

Remove vpath2 from the VG by issuing the reducevg command. When the process is
complete, verify that the source ESS LUN is removed from the VG by using the lsvg -p
command (Example 8-68).
Example 8-68 Removing and verifying the LUN from the VG

# reducevg chuckvg vpath2
# lsvg -p chuckvg
chuckvg:
vpath3 active 74 1 00..00..00..00..01
vpath4 active 14 1 00..00..00..00..01
vpath5 active 79 5 00..00..00..00..05
vpath6 active 15 15 03..03..03..03..03
vpath7 active 79 79 16..16..15..16..16
Check the vpaths and associated hdisks by running the lsvpcfg command as shown in
Example 8-69.
Example 8-69 Results of running the lsvpcfg command

# lsvpcfg
vpath2 (Avail pv ) 70022665 = hdisk25 (Avail ) hdisk2 (Avail ) hdisk3 (Avail )
hdisk4 (Avail )
hdisk35 (Avail )
hdisk36 (Avail )
vpath5 (Avail pv chuckvg) 75L47414005 = hdisk14 (Avail ) hdisk17 (Avail ) hdisk29
vpath7 (Avail pv chuckvg) 75L47414004 = hdisk16 (Avail ) hdisk28 (Avail) hdisk37
(Avail ) hdisk38
(Avail)

Deleting the device definitions from the host ODM
Remove the device definitions for all the source ESS vpaths and all their associated hdisk
paths. For example, to remove vpath2 and its associated hdisk paths, issue the rmdev
commands as shown in Example 8-70.
Example 8-70 Deleting the device definitions

rmdev -dl vpath2
rmdev -dl hdisk19
rmdev -dl hdisk22
rmdev -dl hdisk25
In Example 8-70, note that only vpath2 and its associated disks were deleted. Use the same
process to delete all the vpaths and the associated hdisks that are owned by vpath3 and
vpath4.
Verifying that the device definitions are removed

Verify that there are no defined disks any more by issuing the lsdev -Cc disk command as
Example 8-71 Results from the lsdev -Cc disk command

# lsdev -Cc disk
hdisk0 Available 1S-08-00-4,0 16 Bit LVD SCSI Disk Drive
hdisk14 Available 1V-08-01 IBM FC 2107
hdisk16 Available 1H-08-01 IBM FC 2107
vpath5 Available Data Path Optimizer Pseudo Device Driver
Removing the source zone definition

Remove the source zone definition from the switch fabric.

Unassigning the source LUNs from the host server
Unassign the source ESS volumes from the storage unit and verify that the ESS volumes are
no longer showing on the system. See Figure 8-25.
Figure 8-25 Unassigning ESS LUNs
8.7.4 High-level migration plan using AIX LVM mklvcopy and syncvg
commands
The AIX LVM is flexible. Under certain constraints or requirements for data spreading or
consolidation across LUNs, it might be easier to use the mklvcopy and syncvg commands. If
you choose to use the mklvcopy command, check that there is enough space in the VG on the
DS8000 LUNs collectively to create the mirrors. Table 8-10 shows the sequence, commands,
and a brief explanation of this scenario.
Table 8-10 Example migration procedure using mklvcopy
Step 1. Identify the ESS source lsvpcfg Gather information about the
LUNs. AIX host, such as the number of
ESS LUNs and sizes.
Step 2. Assign the DS8000 chvolgrp -dev image_id On the DS8000, use the
LUNs to the host. -action add-volume vol_# V3 DS8000 CLI (dscli) to assign
the LUNs. Where image_id is
the DS8000 ID and vol_# is the
number of the DS8000 volume
in hex.
Step 3. Discover the DS8000 cfgmgr On the AIX host, discover the
LUNs. newly assigned LUNs.

Step 4. Determine the lsvpcfg This command works using the

difference between the ESS SDD driver on the AIX host. The
and DS8000 LUNs. source LUNs are ESS and the
target LUNs are DS8000. Look
at the serial numbers of the
LUNs in the output.
Step 5. Identify the sizes of the bootinfo -s vpath# Where # is the number of the
DS8000 target LUNs. vpath.
Step 6. Move the DS8000 extendvg vg-name vpath# Where vg_name is the name of
LUNs into the VGs the VG and # is the number of
appropriately. the vpath.
LUNs are added to the VG. the VG.
Step 8. Determine how the LVs lslv -l lv_name Where lv_name is the name of
are spread across the vpaths. the LV.
Step 9. Reserve free space on mklv -y lvdummy vg_name PPs Where vg_name is the name of
each LUN for an even spread of vpath# the VG and # is the number of
the data across LUNs. the vpath.
Step 10. Copy LV data from the mklvcopy lv_name 2 vpath# Where lv_name is the name of
ESS source LUNs to the vpath# the LV and # is the number of
DS8000 target LUNs. the vpath of the target DS8000
LUN.
Step 11. Verify that the LV lslv -l lv_name Where lv_name is the name of
copies are correct. the LV.
Step 12. Synchronize the LV syncvg -v vg_name Where vg_name is the name of
data from the ESS source LUNs the VG.
to the DS8000 target LUNs.
Step 13, Verify that the sync is lsvg -l vg_name If the lv still shows stale, you
displayed as sync’d rather than need to resync it before
stale. proceeding.
Step 14. Verify the source and lslv -l lv_name Where lv_name is the name of
target LUNs for each LV. the LV.
Step 15. Remove the source rmlvcopy lv_name 1 vpath# Where lv_name is the name of
copy of the LV from the ESS the LV and # is the number of
LUNs. the vpath of the target DS8000
LUN.
Step 16. Verify that all the lsvg -p vg_name Where vg_name is the name of
source ESS LUNs are free with the VG.
no data.
Step 17. Remove the ESS reducevg vg_name vpath# Where vg_name is the name of
source LUNs from the VGs and lsvg -p vg_name the VG and # is the number of
verify that the source ESS the vpath.
LUNs are removed from the
VG.
Step 18. Delete the device rmdev -dl vpath# Where # is the number of the
definitions from the host ODM. rmdev -dl hdisk# vpath and appropriate hdisk.

Step 19. Verify that the device lsdev -Cc disk Check that there are no defined
definitions are removed. disks.
Step 20. Remove the source n/a Removes data path between
zone definition from the switch AIX host and disk subsystem
fabric.
Step 21. In the ESS, unassign 1. Click Modify Volume This sequence is done from the
the LUNs from the host server. Assignments. ESS Specialist GUI.
2. In the Volume assignments
window, select First in the
Host Nicknames list.
3. Click Perform Sort.
4. Press the Ctrl key and
select the first volume
assigned to the host.
5. In the Action list, select
Unassign selected
volume(s) to target hosts.
6. In the Target Hosts list,
select all targeted volumes
using the Ctrl key.
7. Click Perform
Configuration Update.
8. Click OK.
8.7.5 Detailed migration steps using mklvcopy and syncvg

The migration procedure has the following detailed steps, using an AIX V5.2 with both the
2105 and the 2107 SDD drivers:
򐂰 Identifying the source ESS LUNs in the targeted VGs
򐂰 Assigning the DS8000 LUNs to the AIX host
򐂰 Assigning and discover the DS8000 LUNs
򐂰 Determining the differences between the ESS and DS8000 LUN
򐂰 Identifying the sizes of the DS8000 LUNs
򐂰 Moving the DS8000 LUNs into the VGs
򐂰 Verifying that the vpaths are correctly added to the VG
򐂰 Determining how the LVs are spread across the vpaths
򐂰 Reserving free space for an even spread of the data
򐂰 Copying LV data from source LUNs to target LUNs
򐂰 Verifying that the copies are correct
򐂰 Synchronizing the copies
򐂰 Checking for stale partitions
򐂰 Verifying LV copies
򐂰 Removing the source copies from the ESS LUNs
򐂰 Verifying that all the source ESS LUNs are free with no data
򐂰 Removing the ESS source LUNs from the VGs
򐂰 Deleting the device definitions from the host ODM
򐂰 Removing the source zone definition
򐂰 Unassigning the LUNs from the host server in the ESS

Identifying the source ESS LUNs in the targeted VGs
Issue the lsvg -p command against the target vg_name to determine the size of the LUNs
that you need to assign from the DS8000 (Example 8-72). To identify the sizes of the LUNs in
the targeted VG, multiply the number of PPs for the vpath by the PP size of the VG. This
calculation gives you the LUN size in MB. The formula is:
(LUNPPs) * vgPPsize = LUN size
If you take the Total PPs for vpath2 and multiply it by the VG PP size of 64 you get 4736 MB:
(74) * (64) = 4736 MB
Example 8-72 Results from running the lsvg -p command

# lsvg -p chuckvg
chuckvg:
vpath2 active 74 0 00..00..00..00..00
vpath3 active 74 1 00..00..00..00..01
vpath4 active 14 1 00..00..00..00..01
Assigning the DS8000 LUNs to the AIX host

Use the dscli with the following commands to assign the new volumes (4004 - 4006 to V3),
and initiate the host connections in the DS8000 (Example 8-73). The volumes and volgrp
were previously created.
Example 8-73 Using dscli commands to assign storage to the AIX host
mkfbvol -dev IBM.2107-75L4741 -extpool P4 -name 8300_#h -type ds -cap 5 4004-4005
mkfbvol -dev IBM.2107-75L4741 -extpool P4 -name 8300_#h -type ds -cap 1 4006
mkvolgrp -dev IBM.2107-75L4741 IBM.2107-7581981 scsimask -volume 4004-4006 V3
mkhostconnect -dev IBM.2107-75L4741 -wwname 10000000C93E007C -hosttype pSeries
mkhostconnect -dev IBM.2107-75L4741 -wwname 10000000C93E0059 -hosttype pSeries
Issue the showvolgrp command as shown in Example 8-74 to verify that LUNs are assigned
to V3.
Example 8-74 Displaying LUNs in a volgrp using the dscli

IBM.2107-75L4741
Name p615_tic6_vg
ID V3
Type SCSI Mask
Vols 4004 4005 4006
Notice that LUNs 4004-4005 are assigned to volgrp V3.

Assigning and discover the DS8000 LUNs
Assign the DS8000 LUNs and run cfgmgr on the AIX host system.
Determining the differences between the ESS and DS8000 LUN

You can view the differences between the ESS and DS8000 by running the lsvpcfg
command. Identify the DS8000 and ESS LUNs by the LUN ID displayed in the output of the
lsvpcfg command as shown in Example 8-75.
Example 8-75 Identifying ESS and DS8000 LUNs

# lsvpcfg
vpath2 (Avail pv chuckvg) 70022665 = hdisk19 (Avail ) hdisk22 (Avail ) hdisk25 (Avail )
hdisk34 (Avail )
hdisk35 (Avail )
hdisk36 (Avail )
vpath5 (Avail ) 75L47414005 = hdisk14 (Avail ) hdisk17 (Avail ) hdisk29 (Avail ) hdisk32
(Avail )
vpath6 (Avail ) 75L47414006 = hdisk15 (Avail ) hdisk18 (Avail ) hdisk30 (Avail ) hdisk33
(Avail )
vpath7 (Avail pv ) 75L47414004 = hdisk16 (Avail ) hdisk28 (Avail ) hdisk37 (Avail ) hdisk38
(Avail )
The first three digits of the ESS LUN IDs are the LUN hex ID, and the last five digits are the
ESS serial number. The DS8000 LUNs are the opposite: the first eight digits are the DS8000
Serial number, and the last four digits are the LUN hex ID. For example, because vpath 5 is a
DS8000 LUN, 75L4741 is the serial number of the unit and 4005 is the LUN ID. The 40 is the
LSS and 05 is the LUN ID in hex.
Identifying the sizes of the DS8000 LUNs

You need to know the sizes of the DS8000 LUNs to move them appropriately into the volume
groups (VGs). If you use the mklvcopy command, make sure that there is enough space in the
VG on the DS8000 LUNs collectively to create the mirrors. Issue the bootinfo -s
vpath/hdisk command to determine the sizes in MB as shown in Example 8-76.
Example 8-76 Output showing the results of the bootinfo command

5120
1024
5120
Moving the DS8000 LUNs into the VGs

Issue the extendvg command to move the DS8000 LUNs into the targeted volume groups as
Example 8-77 Using the extendvg command to move LUNs into a VG

# extendvg chuckvg vpath5 vpath6 vpath7
0516-1254 extendvg: Changing the PVID in the ODM.

Verifying that the vpaths are correctly added to the VG
Verify that the vpaths are correct with the lsvg -p command as shown in Example 8-78.

# lsvg -p chuckvg
chuckvg:
vpath2 active 74 0 00..00..00..00..00
vpath3 active 74 1 00..00..00..00..01
vpath4 active 14 1 00..00..00..00..01
vpath5 active 79 79 16..16..15..16..16
vpath6 active 15 15 03..03..03..03..03
vpath7 active 79 79 16..16..15..16..16
Determining how the LVs are spread across the vpaths

It is important to know whether you have special host LUN spreading requirements. Issue the
lslv -l command to obtain this information as shown in Example 8-79.
Example 8-79 Using the lslv -l command to check for host LUN spreading requirements
# lslv -l ess5GBlv1
ess5GBlv1:/ess5GB1
PV COPIES IN BAND DISTRIBUTION
vpath0 037:000:000 100% 015:000:000:007:015
vpath1 036:000:000 100% 007:000:000:015:014
# lslv -l ess5GBlv2
ess5GBlv2:/ess5GB2
vpath1 038:000:000 100% 008:015:014:000:001
vpath0 030:000:000 100% 000:008:014:008:000
vpath2 005:000:000 100% 001:000:000:003:001
# lslv -l ess1GBlv
ess1GBlv:/ess1GB
vpath2 009:000:000 100% 002:003:002:000:002
vpath0 004:000:000 100% 000:004:000:000:000
# lslv -l loglv00
loglv00:N/A
vpath0 001:000:000 100% 000:001:000:000:000
Reserving free space for an even spread of the data

Reserve free space on each LUN for an even spread of the data across LUNs. To ensure an
even spread of the LV across LUNs in the VG, reserve free space on each LUN. Reserve the
space by making a dummy LV on each of the targeted DS8000 LUNs using the mklv
Example 8-80 Making a dummy LV

# mklv -y lvdummy chuckvg 39 vpath5
lvdummy
# extendlv lvdummy 39 vpath7

Verify that the dummy LV has reserved the space on the targeted LUNs by issuing the
lsvg -p command as shown in Example 8-81.
Example 8-81 Output from the lsvg -p command

# lsvg -p chuckvg
chuckvg:
vpath0 active 74 2 00..02..00..00..00
vpath1 active 74 0 00..00..00..00..00
vpath2 active 14 0 00..00..00..00..00
vpath5 active 79 37 16..00..00..05..16
vpath6 active 15 15 03..03..03..03..03
vpath7 active 79 37 05..00..00..16..16
Copying LV data from source LUNs to target LUNs

Copy LV data from the ESS source LUNs to the DS8000 target LUNs using the mklvcopy
command shown in Example 8-82.
Example 8-82 Using the mklvcopy command

# mklvcopy ess5GBlv1 2 vpath5 vpath7
Verifying that the copies are correct

Verify the that LV copies are correctly made by issuing the lslv -l command as shown in
Example 8-83.
Example 8-83 Verifying the copies with the lslv -l command

# lslv -l ess5GBlv1
ess5GBlv1:/ess5GB1
vpath0 037:000:000 100% 015:000:000:007:015
vpath5 037:000:000 100% 016:000:000:005:016
vpath1 036:000:000 100% 007:000:000:015:014
vpath7 036:000:000 100% 005:000:000:016:015
Remove lvdummy and continue making the other LV copies by using the rmlv command as
Example 8-84 Using the rmlv command to remove lvdummy

# rmlv lvdummy
Warning, all data contained on logical volume lvdummy will be destroyed.
rmlv: Do you wish to continue? y(es) n(o)? y
rmlv: Logical volume lvdummy is removed.
Synchronizing the copies

Synchronize the copies by issuing the syncvg -v command as shown in Example 8-85.
Example 8-85 Using the sync -v command

# syncvg -v chuckvg

Checking for stale partitions
Verify that there are no stale partitions by issuing the lsvg -l command as shown in
Example 8-86.
Example 8-86 Using the lsvg -l command to verify the PP state

lsvg -l chuckvg
chuckvg:
ess1GBlv jfs2 13 26 3 open/syncd /ess1GB
Verifying LV copies
Verify that LV copies are shown from both the source and target LUNs for each LV by issuing
the lslv -l command (Example 8-87).
Example 8-87 Using the lslv -l command to verify lv copies on the source and target LUNs
# lslv -l ess5GBlv1
ess5GBlv1:/ess5GB1
vpath0 037:000:000 100% 015:000:000:007:015
vpath5 037:000:000 100% 016:000:000:005:016
vpath1 036:000:000 100% 007:000:000:015:014
vpath7 036:000:000 100% 005:000:000:016:015
# lslv -l ess5GBlv2
ess5GBlv2:/ess5GB2
vpath1 038:000:000 100% 008:015:014:000:001
vpath7 037:000:000 100% 011:010:015:000:001
vpath5 036:000:000 100% 000:010:015:011:000
vpath0 030:000:000 100% 000:008:014:008:000
vpath2 005:000:000 100% 001:000:000:003:001
# lslv -l ess1GBlv
ess1GBlv:/ess1GB
vpath2 009:000:000 100% 002:003:002:000:002
vpath6 013:000:000 100% 003:003:003:003:001
vpath0 004:000:000 100% 000:004:000:000:000
# lslv -l loglv00
loglv00:N/A
vpath0 001:000:000 100% 000:001:000:000:000
vpath5 001:000:000 100% 000:001:000:000:000
Removing the source copies from the ESS LUNs

To remove the source copies from the ESS LUNs, issue the rmlvcopy command as shown in
Example 8-88.
Example 8-88 Using the rmlvcopy command to remove the source ESS source LUNs
# rmlvcopy ess5GBlv1 1 vpath0 vpath1
# rmlvcopy ess5GBlv2 1 vpath0 vpath1 vpath2

# rmlvcopy ess1GBlv 1 vpath0 vpath2
# rmlvcopy loglv00 1 vpath0
Verifying that all the source ESS LUNs are free with no data
Verify that all the source ESS LUNs are free with no data by checking the LUNs in the VG.
Issue the lsvg -p command as shown in Example 8-89.
Example 8-89 Output results from the lsvg -p command

# lsvg -p chuckvg
chuckvg:
vpath0 active 74 74 15..15..14..15..15
vpath1 active 74 74 15..15..14..15..15
vpath2 active 14 14 03..03..02..03..03
vpath5 active 79 5 00..05..00..00..00
vpath6 active 15 2 00..00..00..00..02
vpath7 active 79 6 00..06..00..00..00
Removing the ESS source LUNs from the VGs

Remove vpath2 from the VG by issuing the reducevg command. Then verify that the source
ESS LUN is removed from the VG by using the lsvg -p command as shown in
Example 8-90.
Example 8-90 An example of removing and verifying the LUN from the VG
# reducevg chuckvg vpath2
# lsvg -p chuckvg
chuckvg:
vpath3 active 74 1 00..00..00..00..01
vpath4 active 14 1 00..00..00..00..01
vpath5 active 79 5 00..00..00..00..05
vpath6 active 15 15 03..03..03..03..03
vpath7 active 79 79 16..16..15..16..16
You can check the vpaths and associated hdisks by running the lsvpcfg command as shown
in Example 8-91.
Example 8-91 Results of running the lsvpcfg command

# lsvpcfg
hdisk4 (Avail )
hdisk35 (Avail )
hdisk36 (Avail )

vpath7 (Avail pv chuckvg) 75L47414004 = hdisk16 (Avail ) hdisk28 (Avail) hdisk37
(Avail ) hdisk38
(Avail)
Deleting the device definitions from the host ODM

Remove the device definitions for all the source ESS vpaths and all their associated hdisk
paths. For example, to remove vpath2 and its associated hdisk paths, issue the rmdev
Example 8-92 Deleting the device definitions

rmdev -dl vpath2
rmdev -dl hdisk19
rmdev -dl hdisk22
rmdev -dl hdisk25
In Example 8-92, only vpath2 and its associated disks were deleted. Use the same process to
delete all the vpaths and the associated hdisks owned by vpath3 and vpath4.

Verify that there are no defined disks left by issuing the lsdev -Cc disk command
(Example 8-93).
Example 8-93 Results from issuing the lsdev -Cc disk command
# lsdev -Cc disk
Removing the source zone definition

Remove the source zone definition from the switch fabric.

Unassigning the LUNs from the host server in the ESS
Unassign the source ESS volumes from the storage unit and verify that the ESS volumes are
no longer showing on the system (Figure 8-26).
Figure 8-26 Unassigning ESS LUNs
8.8 Data migration in a PowerHA clustered environment using

AIX LVM mirroring
This section addresses data migration using AIX Logical Volume Manager (LVM) to migrate
data from one set of IBM DS4000 LUNs to another in a PowerHA clustered environment. In
this example, a file system created on a logical volume striped across four physical LUNs is
migrated to a set of two physical LUNs using AIX LVM mirroring. The method for mirroring
logical volumes (LV) in a clustered environment is similar to a stand-alone AIX environment.
However, there are significant differences. The main difference is the way a volume group is
used. In a stand-alone configuration, the volume group is a dedicated resource, whereas in a
clustered configuration it is shared.
A key element of any PowerHA cluster is the data used by the highly available applications.
This data is stored on AIX LVM entities. PowerHA clusters use the capabilities of the LVM to
make this data accessible to multiple nodes. When you change the definition of a shared LVM
component in a cluster, the operation updates the following items:
򐂰 The LVM data that describes the component on the local node
򐂰 The Volume Group Descriptor Area (VGDA) on the disks in the volume group
AIX LVM enhancements allow all nodes in the cluster to be aware of changes to a volume
group, logical volume, and file system when those changes are made. Depending on the
version of AIX and PowerHA software, LVM changes might not propagate automatically to all
cluster nodes. Verify that the LVM definition of a volume group is the same on all cluster
nodes before a failover occurs.

To change the LVM configuration on a stand-alone AIX server, you normally use AIX
commands such as extendvg and mklvcopy. A shared resource in a PowerHA cluster can be
reliably controlled by only one resource manager at a time. For resources defined as part of
PowerHA resource groups, PowerHA must be the only resource manager controlling the
resource.
Refrain from using AIX commands to modify such resources, and use only PowerHA
operations on the resources. Use only Cluster Single Point Of Control (C-SPOC) operations
on shared volume groups, physical disks, and file systems when the cluster is active. Using
AIX commands on shared volume groups while the cluster is active can result in the volume
group becoming inaccessible, and can corrupt data.
8.8.1 PowerHA test environment

The example PowerHA environment contains a dual-node PowerHA cluster with a DS4000
used as a shared disk subsystem. Four LUNs are presented to both cluster nodes. A volume
group is created using four hdisks. The logical volume is striped across all available hdisks,
and the JFS2 file system is created on the logical volume as shown in Figure 8-27.
ITSO
PowerHA CLUSTER Diagram
Cluster Name = itsoclstr
HA Network : net_ether_01
Gateway
ent1
ent2
ent0
ent1
9.53.11.1
(255.255.252.0)
en0 en1 en0 en1
Boot IP Standby IP Boot IP Standby IP

itso_engine1_boot itso_engine1_stdby itso_engine2_boot itso_engine2_stdby
9.53.11.227 192.168.2.10 9.53.11.228 192.168.2.11
Service IP Service IP
itso_cluster_svc itso_cluster_svc
9.53.11.226 9.53.11.226
itso_engine1 (Priority. 1) itso_engine2 (Priority. 2)

Service IP : itso_cluster_svc (9.53.11.226) Startup Policy: Online Using Distribution Policy
Service IP : itso_cluster_svc (9.53.11.226)
VolumeGroup : itsovg Fallover Policy: Fallover to Next Priority node VolumeGroup : itsovg
Resource Group: itso_rg in List (FONP) Resource Group: itso_rg
Fallback Policy: Never Fallback
itso_engine1 itso_engine2
fcs0
fcs2
fcs0
fcs2
(p520 AIX physical) (p520 AIX physical)
SAN Switch SAN Switch
Running Software
AIX 6100-06-03-1048
itsovg PowerHA 6.1.0 SP5
itsolv
hdisk2
hdisk3
hdisk4
hdisk5
Disk Heartbeat
net_diskhb_01 net_diskhb_02
itso_engine1: /dev/hdisk2 itso_engine1: /dev/hdisk3
IBM DS4000 itso_engine2: /dev/hdisk2 itso_engine2: /dev/hdisk3
Figure 8-27 Data migration test PowerHA environment

You can get information about the DS4000 LUNs using the mpio_get_config -Av command
as shown in Example 8-94.
Example 8-94 IBM DS4000 LUNs

# mpio_get_config -Av
Frame id 0:
Storage Subsystem worldwide name: 60ab80026806c000049d5867c
Controller count: 2
Partition count: 1
Partition 0:
Storage Subsystem Name = 'DRS_FAStT1'
hdisk LUN # Ownership User Label
hdisk2 0 A (preferred) ITSO_vol_s001
hdisk4 9 B (preferred) ITSO_vol_s003
Similarly, you can get information about the LVM entity definitions using the lsvg itsovg
Example 8-95 LVM entity definitions

# lsvg itsovg
VOLUME GROUP: itsovg VG IDENTIFIER: 0004afd80000d70000000131205383d4
VG STATE: active PP SIZE: 32 megabyte(s)
VG PERMISSION: read/write TOTAL PPs: 2044 (65408
megabytes)
MAX LVs: 256 FREE PPs: 2032 (65024
megabytes)
LVs: 1 USED PPs: 12 (384 megabytes)
OPEN LVs: 1 QUORUM: 3 (Enabled)
TOTAL PVs: 4 VG DESCRIPTORS: 4
STALE PVs: 0 STALE PPs: 0
ACTIVE PVs: 4 AUTO ON: no
Concurrent: Enhanced-Capable Auto-Concurrent: Disabled
VG Mode: Concurrent
Node ID: 3 Active Nodes:
MAX PPs per VG: 32512
MAX PPs per PV: 1016 MAX PVs: 32
LTG size (Dynamic): 256 kilobyte(s) AUTO SYNC: no
HOT SPARE: no BB POLICY: relocatable
PV RESTRICTION: none
# lsvg -l itsovg
itsovg:
itsolv jfs2 12 12 4 open/syncd /itsofs
# lslv itsolv
LLOGICAL VOLUME: itsolv VOLUME GROUP: itsovg
LV IDENTIFIER: 0004afd80000d70000000131205383d4.1 PERMISSION: read/write
VG STATE: active/complete LV STATE: opened/syncd
TYPE: jfs2 WRITE VERIFY: off
MAX LPs: 512 PP SIZE: 32 megabyte(s)
COPIES: 1 SCHED POLICY: striped

LPs: 12 PPs: 12
STALE PPs: 0 BB POLICY: relocatable
INTER-POLICY: maximum RELOCATABLE: no
INTRA-POLICY: middle UPPER BOUND: 4
MOUNT POINT: /itsofs LABEL: /itsofs
MIRROR WRITE CONSISTENCY: on/ACTIVE
EACH LP COPY ON A SEPARATE PV ?: yes (superstrict)
Serialize IO ?: NO
STRIPE WIDTH: 4
STRIPE SIZE: 64K
# lslv -m itsolv
itsolv:/itsofs
LP PP1 PV1 PP2 PV2 PP3 PV3
0001 0104 hdisk2
0002 0104 hdisk3
0003 0104 hdisk4
0004 0104 hdisk5
0005 0105 hdisk2
0006 0105 hdisk3
0007 0105 hdisk4
0008 0105 hdisk5
0009 0106 hdisk2
0010 0106 hdisk3
0011 0106 hdisk4
0012 0106 hdisk5
0032 0111 hdisk5
# lsfs /itsofs
Name Nodename Mount Pt VFS Size Options Auto
Accounting
/dev/itsolv -- /itsofs jfs2 786432 rw no no
Use the /usr/es/sbin/cluster/utilities/cltopinfo -n command to get information about

PowerHA topology and resource group definitions as shown in Example 8-96.
Example 8-96 PowerHA topology and resource group definitions

# /usr/es/sbin/cluster/utilities/cltopinfo -n
NODE itso_engine1:
Network net_diskhb_01
itso_engine1_hdisk2_01 /dev/hdisk2
Network net_ether_01
itso_cluster_svc 9.53.11.226
itso_engine1_boot 9.53.11.227
NODE itso_engine2:

# /usr/es/sbin/cluster/utilities/clshowres -g'itso_rg'
Resource Group Name itso_rg

Participating Node Name(s) itso_engine1 itso_engine2
Startup Policy Online On Home Node Only
Fallover Policy Fallover To Next Priority Node
In The List
Fallback Policy Never Fallback
Site Relationship ignore
Dynamic Node Priority
Service IP Label itso_cluster_svc
Filesystems /itsofs
Filesystems Consistency Check fsck
Filesystems Recovery Method sequential
...
Volume Groups itsovg
Concurrent Volume Groups
Use forced varyon for volume groups, if necessary false
Automatically Import Volume Groups false
...
To migrate data from the current set of hdisks, you need to migrate a single logical volume
(LV), itsolv, containing /itsofs file system. In the example, the second set of LUNs is
allocated on the same IBM DS5000 storage subsystem. You do not need to replace or
upgrade the multipathing device driver.
When both sets of LUNs are discovered on both cluster nodes, use C-SPOC commands to
mirror the volume group between both sets of LUNs. After the volume group is mirrored, you
can remove the mirrored copy from the original set of LUNs and remove those LUNs from the
volume group.
In the example, you also perform consolidation by reducing the number of LUNs. The data
from the original set of four LUNs are migrated to a set of two LUNs. The capacity of the
target LUNs (32 GB) is twice the capacity of the source LUNs (16 GB).
In the example, the logical volume is striped across all available volumes within the volume
group. Logical volume striping is used to improve performance by balancing the workload
between physical devices. As you consolidate the data on two volumes, the width of striping is
reduced from four volumes to two volumes. Verify the version of AIX you are working with
before using this function because LVM striping width reduction is OS version dependent.

8.8.2 High-level migration plan using PowerHA C-SPOC LVM mirroring
Table 8-11 shows the high-level steps for a data migration using C-SPOC LVM mirroring.
Table 8-11 Example migration procedure using PowerHA C-SPOC LVM mirroring
Step1. Identify the DS5000 mpio_get_config -Av Gather information about the
source LUNs and LUN sizes. bootinfo -s hdiskX AIX host including the number
of DS5000 LUNs and sizes.
Step 2. Mapping of new set of DS5000 Storage Manager GUI Use DS5000 Storage Manager
DS5000 LUNs to the cluster to map new set of LUNs to the
nodes. cluster nodes.
Step 3. Discover new DS5000 cfgmgr On the cluster nodes, discover

LUNs. the newly assigned LUNs.
Step 4. Determine LUN sizes mpio_get_config -Av Where # is the number of the
for both sets of LUNs bootinfo -s hdisk# hdisk.
Step 5. Include the new LUNs smitty cl_extendvg Use the C-SPOC smitty panel
into the existing volume group to assign a PVID to a new
volume and extend the volume
group using the new volume.
Must be repeated for all new
volumes.
LUNs are added to the VG on the VG
all cluster nodes.
Step 7. Identify the logical lsvg -l vg_name Where vg_name is the name of
volumes (LVs) to migrate. the VG.
Step 8. Create LV copies on the smitty cl_mirrorvg Use the C-SPOC smitty panel
new set of volumes. to create an LV mirror copy.
Step 9. Synchronize LVM smitty cl_syncvg_lv Use the C-SPOC smitty panel
mirrors. to synchronize LVM mirrors by
LV.
Step 10. Monitor the lsvg vg_name Where vg_name is the name of
synchronization process. the VG.
Step 11. Verify LVM mirror lsvg vg_name Where vg_name and lv_name
consistency. lslv lv_name are the names of VG and
mirscan -l itsolv mirrored LV.
Step 12. Remove LV copies smitty cl_unmirrorvg Use the C-SPOC smitty panel
from the old set of volumes. to unmirror VG.
Step 13. Remove old volumes smitty cl_reducevg Use the C-SPOC smitty panel
from the VG. lsvg -p vg_name for cluster-wide removal of old
volume from VG. The lsvg
command must be run on each
cluster node for verification.
Step 14. Replace smitty Use the C-SPOC smitty panel

communication devices in cm_change_show_communicatio to replace original heartbeat
PowerHA heartbeat networks. n_interfaces_devices.select device with the devices residing
on the new set of LUNs

Step 15. Delete the device rmdev -dl hdisk# Where # is the number of the
definitions from the server vpath and the appropriate
ODM. hdisk.
Step 16. Verify that the device mpio_get_config -Av Check that there are no defined
definitions are removed. lsdev -Cc disk hdisks.
Step 17. Unassign old DS5000 DS5000 Storage Manager GUI Use DS5000 Storage Manager
volumes from all cluster nodes. to remove mapping of old set of
LUNs to the cluster nodes.
8.8.3 Detailed migration steps using PowerHA C-SPOC LVM mirroring

Migrating data from one set of LUNs to another in PowerHA clustered environments has the
following detailed steps. In this example, no new device drives or multipathing software are
needed because the new LUN allocation is done from the same disk subsystem.
򐂰 Identifying the DS5000 source LUNs and LUN sizes
򐂰 Mapping the new set of DS5000 LUNs to the cluster nodes
򐂰 Discovering new DS5000 LUNs
򐂰 Determining LUN sizes for both sets of LUNs
򐂰 Including the new LUNs into the existing volume group
򐂰 Verifying that the DS5000 LUNs are added to the VG
򐂰 Identifying LVs to migrate
򐂰 Creating LV copies on the new set of volumes
򐂰 Synchronizing LVM mirrors
򐂰 Monitoring the synchronization process
򐂰 Verifying LVM mirror consistency
򐂰 Removing LV copies from the old set of volumes
򐂰 Removing old volumes from the VG
򐂰 Replacing communication devices in PowerHA heartbeat networks
򐂰 Deleting the device definitions from the server ODM
򐂰 Unassigning old set of DS5000 LUNs from the cluster nodes
Identifying the DS5000 source LUNs and LUN sizes

The AIX command used to identify DS5000 LUNs depends on the firmware level of the
DS5000. In the example, the firmware level of the DS5000 requires the use of MPIO
multipathing on the server side. The mpio_get_config command shows information about the
MPIO-based DS5000 subsystem and the hdisks associated with it (Example 8-97).
Specifically, the command shows the following information about the subsystem:
򐂰 The assigned name of the subsystem
򐂰 The worldwide name of the subsystem
򐂰 A list of hdisks in the Available state that are associated with the subsystem
Example 8-97 DS5000 LUN identification

Frame id 0:
Controller count: 2
Partition count: 1
Partition 0:

To determine the size of hdisks associated with DS5000 LUNs, use LVM commands that
report the size of hdisks such as lspv and lsvg. These commands work when a hdisk is
already included into a VG (lsvg) or has a Physical Volume Identifier (PV ID) assigned. If no
PV ID is assigned to the hdisk, use the bootinfo command to determine the size. The
bootinfo command reports the size of a hdisk in Megabytes (MB) as shown in Example 8-98
on page 356.

# for i in 2 3 4 5 ;do a=`bootinfo -s hdisk$i` ;echo hdisk$i capacity $a MB; done
hdisk2 capacity 16384 MB
Mapping the new set of DS5000 LUNs to the cluster nodes

In this example, the Default group host group is used as the cluster nodes, which are the only
servers zoned to DS5000. The DS5000 Storage Manager GUI window is used for LUN
mapping (Figure 8-28).
Figure 8-28 Mapping DS5000 LUNs to cluster nodes

Discovering new DS5000 LUNs
After all new DS5000 LUNs are mapped, run cfgmgr on all cluster nodes to discover the newly
assigned LUNs.
Determining LUN sizes for both sets of LUNs

When LUNs are created on a DS5000 disk subsystem, each LUN gets assigned a name. Use
a naming convention that helps you to identify sets of LUNs. In this example, the s character
after the second underscore in ITSO_vol_s00# indicates the source set. Similarly, the t
character in ITSO_vol_t00# indicates the target set (Example 8-99).
Example 8-99 Identifying DS5000 LUNs

# /usr/bin/mpio_get_config -Av
Frame id 0:
Controller count: 2
Partition count: 1
Partition 0:
hdisk6 11 B (preferred) ITSO_vol_t001
hdisk7 12 A (preferred) ITSO_vol_t002
The newly discovered hdisk6 and hdisk7 volumes do not have PVIDs assigned yet. To verify
the LUN sizes use the bootinfo command (Example 8-100).

# for i in 2 3 4 5 6 7;do a=`bootinfo -s hdisk$i` ;echo hdisk$i capacity $a MB;
done
The capacity of the newly discovered hdisk6 and hdisk7 volumes is 32 GB, twice the capacity
of the source volumes. The combined capacity of the new set of volumes matches the
capacity of the old set, 64 GB.

Including the new LUNs into the existing volume group
To extend the existing volume group using the newly discovered volumes, use C-SPOC
commands to maintain consistency across all cluster nodes. The easiest way to start
C-SPOC commands is to use the AIX smitty interface (Example 8-101). Unlike standard AIX
LVM commands, C-SPOC commands require a number of additional parameters such as
resource group name and node names. Although it is possible to add all those parameters in
a command line, it is easier to do through smitty.
Example 8-101 Extending itsovg volume group using C-SPOC smitty interface “smitty cl_extendvg”
+--------------------------------------------------------------------------+
| Select the Volume Group that will hold the new Logical Volume |
| |
| Move cursor to desired item and press Enter. Use arrow keys to scroll. |
| |
| #Volume Group Resource Group Node List |
| itsovg itso_rg itso_engine1,itso_engine2 |
| |
| F1=Help F2=Refresh F3=Cancel |
| F8=Image F10=Exit Enter=Do |
| /=Find n=Find Next |
+--------------------------------------------------------------------------+
+--------------------------------------------------------------------------+
| Physical Volume Names |
| |
| Move cursor to desired item and press Enter. |
| |
| 0004afca24cd59c3 ( hdisk6 on all cluster nodes ) |
| 0004afca24cd5ae2 ( hdisk7 on all cluster nodes ) |
| |
+--------------------------------------------------------------------------+
Add a Volume to a Volume Group
Type or select values in entry fields.

Press Enter AFTER making all desired changes.
[Entry Fields]
VOLUME GROUP name itsovg
Node List itso_engine1,itso_engine2
Reference node itso_engine1
VOLUME names hdisk6
F1=Help F2=Refresh F3=Cancel F4=List

F5=Reset F6=Command F7=Edit F8=Image
F9=Shell F10=Exit Enter=Do

The smitty interface prompts you to select the volume group to expand and the volume to
add to the selected volume group. It then determines the list of cluster nodes and the
reference node. The reference node is the node where the itso_rg resource group is online.
C-SPOC also performs consistency checking along the way.
If the PV IDs do not match, the interface does not allow you to expand the volume group until
the inconsistency is resolved. The simplest resolution is to clear the PVIDs on hdisk6 and
hdisk7 on all cluster nodes using the chdev -l hdisk# -a pv=clear command. You must go
through this process once for every disk used for expanding the volume group.
You can also use command-line interface (CLI) to achieve the same result. However, if you
use the CLI, you must manually check PV ID consistency. The advantage of using the CLI is
that you can automate this task. If you are adding many volumes, CLI can save you time. A
CLI command equivalent to the set of smitty panels shown in Example 8-101 on page 358 is
Example 8-102 C-SPOC CLI command for expanding VG

# /usr/es/sbin/cluster/sbin/cl_extendvg -cspoc -n'itso_engine1,itso_engine2'
-R'itso_engine1' itsovg hdisk6
The C-SPOC command can be run from any node in the cluster, regardless of which node
has the resource group online.
Verifying that the DS5000 LUNs are added to the VG

To verify that the hdisks are added to the VG correctly, run the lsvg -p command on each
cluster node as shown in Example 8-103

<itso_engine1>
# lsvg -p itsovg
itsovg:
hdisk2 active 511 508 103..99..102..102..102
hdisk3 active 511 508 103..99..102..102..102
hdisk4 active 511 508 103..99..102..102..102
hdisk5 active 511 508 103..99..102..102..102
hdisk7 active 1023 1023 205..205..204..204..205
hdisk6 active 1023 1023 205..205..204..204..205
<itso_engine2>
# lsvg -p itsovg
itsovg:
hdisk2 active 511 508 103..99..102..102..102
hdisk3 active 511 508 103..99..102..102..102
hdisk4 active 511 508 103..99..102..102..102
hdisk5 active 511 508 103..99..102..102..102
hdisk7 active 1023 1023 205..205..204..204..205
hdisk6 active 1023 1023 205..205..204..204..205
In this example, both cluster nodes (itso_engine1 and itso_engine2) have the itsovg VG
successfully expanded with the newly discovered hdisk6 and hdisk7.

Identifying LVs to migrate
To identify the LVs to migrate from the source set of LUNs, issue the command lsvg - l on
all cluster nodes as shown in Example 8-104.
Example 8-104 The output results of issuing the lsvg -l command against a VG
<itso_engine1>
# lsvg -l itsovg
itsovg:
itsolv jfs2 12 12 4 open/syncd /itsofs
<itso_engine2>
# lsvg -l itsovg
itsovg:
itsolv jfs2 12 12 4 closed/syncd /itsofs
LV itsolv is in open/syncd state on one of the cluster nodes and closed/syncd on the other.
In PowerHA, cluster JFS2 LV can be opened on only one node at a time. After the file system
is unmounted, the LV switches into closed/syncd state and can be opened by another node.
Attempts to open LV on a cluster node while it is opened on another node results in an error
message. AIX LVM uses locking mechanisms to prevent from more than one node mounting
the same JFS2 file system at a time.
In this example, the JFS2 file system is configured with INLINE JFS2 log. There is no
separate JFS2 log LV device associated with the itsofs file system. If there was a separate
JFS2 log LV, it must be migrated to the new set of LUNs along with the JFS2 LV. By default the
JFS2 file system is configured with a separate jfs2log LV. In this example INLINE jfs2 is used
to keep the configuration simple. In this configuration, you only need to migrate itsolv LV.
Creating LV copies on the new set of volumes

The C-SPOC tool used for mirroring the VG creates an extra copy of all LVs within a VG. You
cannot mirror only some of the LVs in a VG with C-SPOC tools. The standard AIX LVM
commands provide more flexibility for selective mirroring of LVs. Mirroring of all LVs in a VG is
adequate for the task of migrating data from one set of LUNs to another.
To create LV copies of all LVs in the itsovg VG, use the C-SPOC smitty interface. The smitty
interface offers selection choices and performs all the necessary checking on all cluster
nodes before running LVM commands. If you are mirroring data in a few VGs, use the smitty
interface as shown in Example 8-105.
Example 8-105 Mirroring itsovg VG using the C-SPOC smitty interface

+--------------------------------------------------------------------------+
| Select the Volume Group to Mirror |
| |
| |
| |
+--------------------------------------------------------------------------+

+--------------------------------------------------------------------------+
| |
| Move cursor to desired item and press F7. |
| ONE OR MORE items can be selected. |
| Press Enter AFTER making all selections. |
| |
| Auto-select |
| drs_engine1 hdisk2 |
| > drs_engine1 hdisk6 |
| > drs_engine1 hdisk7 |
| |
| F7=Select F8=Image F10=Exit |
| Enter=Do /=Find n=Find Next |
+--------------------------------------------------------------------------+
Mirror a Volume Group

[Entry Fields]
* VOLUME GROUP name itsovg
PHYSICAL VOLUME names hdisk6 hdisk7
Mirror sync mode No Sync +

Number of COPIES of each logical 2 +
partition
Keep Quorum Checking On? no +
Create Exact LV Mapping? no +

After you select the VG to mirror and the set of hdisks to create LV copy on, you are presented
with choices for various attributes. Leave all attributes on the default values except for Mirror
sync mode. Change this attribute value to No Sync. This parameter illustrates the VG
configuration change before data synchronization takes place. In addition, changing this
attribute if you want to increase the parallelism of the data synchronization process. See
“Synchronizing LVM mirrors” on page 363 for more details about the LV synchronization
process.

Verify the results of creating LV copies on the new set of LUNs using the lsvg itsovg
Example 8-106 Results of creating LV copies

# lsvg itsovg
VG PERMISSION: passive-only TOTAL PPs: 4090 (130880 megabytes)
MAX LVs: 256 FREE PPs: 4066 (130112 megabytes)
OPEN LVs: 0 QUORUM: 1 (Disabled)
VG Mode: Concurrent
Node ID: 3 Active Nodes: 1
# lslv itsolv
LOGICAL VOLUME: itsolv VOLUME GROUP: itsovg
VG STATE: active/complete LV STATE: closed/stale
LPs: 12 PPs: 24
Serialize IO ?: NO
STRIPE WIDTH: 4
STRIPE SIZE: 64K
As you can see in Example 8-106 on page 362, you now have a logical volume with two
copies, two stale physical volumes (PVs), and 12 stale physical partitions (PPs). Stale
volumes and stale partitions are present because you delayed the data synchronization. The
AIX command that scans all LV copies and identifies stale PVs is mirscan (Example 8-107).
The mirscan command can run only on a cluster node where the itso_rg resource group is
online.
Example 8-107 Identifying stale PVs

# mirscan -l itsolv
START TIME: Thu Jul 14 00:03:50 CEST:2011
OP STATUS PVNAME PP SYNC IOFAIL LVNAME LP CP TARGETPV
TARGETPP
s SUCCESS hdisk2 104 synced no itsolv 1 1

s FAILURE hdisk7 206 stale no itsolv 2 2
The hdisk6 and hdisk7 physical volumes are in stale state as indicated in the SYNC column.
These PVs contain the second copy of the LV as shown in the CP column. FAILURE in the
STATUS column indicates that the partition was not synchronized and is still incapable of
performing I/O operations because it contains no valid data.
Synchronizing LVM mirrors

Synchronize LV copies using the C-SPOC smitty interface as shown in Example 8-108.
Synchronization can be done either for one LV at a time, or for all LVs in a single VG. In this
example, it makes no difference because there is only one LV to synchronize. A VG-wide
synchronization process, in most cases, is adequate for data migration.
Example 8-108 LV copy synchronization using C-SPOC smitty interface “smitty cl_syncvg_lv”
+--------------------------------------------------------------------------+
| Select the Volume Group that holds the Logical Volume to be Synchronized |
| |
| |
| |
+--------------------------------------------------------------------------+
+--------------------------------------------------------------------------+
| Logical Volume name |
| |
| |

| #itsovg: |
| # LV NAME TYPE LPs PPs PVs LV STATE MO |
| itsolv jfs2 12 24 6 open/stale /i |
| |
+--------------------------------------------------------------------------+
Synchronize LVM Mirrors by Logical Volume

[Entry Fields]
LOGICAL VOLUME name itsolv
* Node List itso_engine1,itso_engine2
Number of Partitions to Sync in Parallel [2] +#

Synchronize All Partitions no +
Delay Writes to VG from other cluster nodes during no +
this Sync

As smitty takes you through the three panels, it prompts you to select the VG and LV to
synchronize, and determines the list of cluster nodes.
In the third smitty panel, select a value for the Number of Partitions to Sync in Parallel
attribute. This attribute specifies the number of Logical Partitions (LPs) that are synchronized
in parallel. The valid range is 1 - 32. The higher the number, the higher the synchronization
rate.
However, consider the performance impact to avoid setting this attribute too high. Find a
balance between acceptable performance impact and the synchronization rate. If you are
unsure, do not change the default attribute value. This value allows the synchronization
process run with minimal performance impact, but it takes the longest to complete.
In this example, the LVM mirror synchronization is started with 2 LPs synchronized in parallel.
The rate depends on the following factors:
򐂰 System type
򐂰 Availability of system resources
򐂰 Number of physical disks in the volume group
򐂰 Performance of the disk subsystem
򐂰 Load on the disk subsystem
򐂰 Many other factors
Understanding the environment is essential in determining the optimal synchronization rate.

Monitoring the synchronization process
To monitor the synchronization process, use the output of lsvg itsovg command. The STALE
PPs field shows the number of PPs that are not synchronized yet. Knowing the size of the PP,
32 MB in the example, you can calculate the capacity that the unsynchronized PPs occupy.
The same method can be used to estimate the duration of the process. Run the command
twice with a few minutes interval between runs. Calculate how many PPs were synchronized
between the runs. Divide the time between the two runs by the number of synchronized PPs
to get the synchronization rate. To estimate the completion time, divide the STALE PPs
number by the calculated rate. For example, if the calculated synchronization rate is 12 PPs
per minute and the number of STALE PPs is 120, it takes approximately 10 minutes to
complete the synchronization.
Verifying LVM mirror consistency

To verify that LVM mirrors are fully synchronized, run the normal AIX LVM commands and
look for stale PVs and PPs in the VG (Example 8-109). If there are remaining stale PVs/PPs,
investigate further.
Example 8-109 Verification of the state of VG and LV

# lsvg itsovg
VG PERMISSION: read/write TOTAL PPs: 4090 (130880 megabytes)
MAX LVs: 256 FREE PPs: 4066 (130112 megabytes)
OPEN LVs: 1 QUORUM: 1 (Disabled)
VG Mode: Concurrent
Node ID: 1 Active Nodes: 3
# lsvg -l itsovg
itsovg:
itsolv jfs2 12 24 6 closed/syncd /itsofs
# lslv itsolv
LOGICAL VOLUME: itsolv VOLUME GROUP: itsovg
VG STATE: active/complete LV STATE: opened/syncd
LPs: 12 PPs: 24

Serialize IO ?: NO
STRIPE WIDTH: 4
STRIPE SIZE: 64K
The results of the verification commands should be identical on all cluster nodes. In the
example, no stale PVs and PPs are found, indicating that the LVM mirrors are fully in sync.
For additional validation, run the mirscan command on the reference node. However, it takes
significantly longer for the mirscan command to complete report generation because it
examines each allocated partition on the specified device.
Example 8-110 shows a mirscan report of fully synchronized LVM mirrors for reference
purposes. The STATUS of physical volumes hdisk6 and hdisk7 is SUCCESS, and SYNC is
synced.
Example 8-110 Examining allocated partitions with mirscan

# mirscan -l itsolv
START TIME: Thu Jul 14 09:06:15 CEST:2011
OP STATUS PVNAME PP SYNC IOFAIL LVNAME LP CP TARGETPV
TARGETPP
END TIME: Thu Jul 14 09:09:13 CEST:2011
Removing LV copies from the old set of volumes

After you confirm that all LV copies are fully synchronized, remove the copy of the LV on the
old set of volumes. This process is sometimes called unmirroring.
The C-SPOC tool used for unmirroring the VG removes an extra copy of all LVs within a VG.
You cannot unmirror only some of the LVs in a VG. In this case, use the standard AIX LVM
commands. For data migration, however, unmirroring of all LVs in a VG is adequate.

Before removing LV copies, consider your backout approach. When the LV is mirrored, you
effectively have multiple (up to three) physical copies of your data. There are two ways you
can “break” LV mirroring. One is to remove and discard the extra copy. The second is to split
that extra copy and form an independent LV. The second approach is addressed in more
detail in 8.8.4, “Migration Backout Considerations” on page 375.
This example uses the first approach of removing and discarding the LV copy on the old set of
volumes: hdisk2 to hdisk5. Use the lslv -m command to view the relationship between PVs
and LV copies as shown in Example 8-111.
Example 8-111 Relationship between LV copies and PVs

# lslv -m itsolv
itsolv:/itsofs
0001 0104 hdisk2 0206 hdisk6
0002 0104 hdisk3 0206 hdisk7
0003 0104 hdisk4 0207 hdisk6
0004 0104 hdisk5 0207 hdisk7
0005 0105 hdisk2 0208 hdisk6
0006 0105 hdisk3 0208 hdisk7
0007 0105 hdisk4 0209 hdisk6
0008 0105 hdisk5 0209 hdisk7
0009 0106 hdisk2 0210 hdisk6
0010 0106 hdisk3 0210 hdisk7
0011 0106 hdisk4 0211 hdisk6
0012 0106 hdisk5 0211 hdisk7
PP1 is the first copy of the LV, PP2 is the second, and PP3 i s the third copy. The first copy of
the LV is on the old set of volumes, hdisk2 to hdisk5 in the PV1 column. Before these
volumes can be excluded from the VG, remove the LV copy on these volumes.
To remove the LV copies of all LVs in itsovg VG, use the C-SPOC smitty interface
(Example 8-112). Smitty offers selection choices and performs all the necessary checking on
all cluster nodes before running LVM commands. If you are unmirroring data on a few VGs,
use the C-SPOC smitty interface to simplify the process.
Example 8-112 Unmirroring itsovg VG using C-SPOC smitty interface “smitty cl_unmirrorvg”
+--------------------------------------------------------------------------+
| Select the Volume Group to Unmirror |
| |
| |
| |
+--------------------------------------------------------------------------+#
+--------------------------------------------------------------------------+
| |

| |
| Auto-select |
| > itso_engine1 hdisk2 |
| itso_engine1 hdisk6 |
| itso_engine1 hdisk7 |
| |
+--------------------------------------------------------------------------+
Unmirror a Volume Group

[Entry Fields]
PHYSICAL VOLUME names hdisk2 hdisk3 hdisk4 hdisk5
Number of COPIES of each logical 1 +

partition

On the second smitty panel, select the set of PVs where the first copy of the LV is located. If
you do not select the PVs, the tool removes the most recent copy of the LV, effectively deleting
the result of the migration. After you confirm the choices on the last smitty panel, the LV copy
is removed. To confirm the LV copy is removed, use lslv command as shown in
Example 8-113.
Example 8-113 Verifying LV copy removal

# lslv -m itsolv
itsolv:/itsofs
0001 0206 hdisk6
0002 0206 hdisk7
0003 0207 hdisk6
0004 0207 hdisk7
0005 0208 hdisk6
0006 0208 hdisk7
0007 0209 hdisk6
0008 0209 hdisk7
0009 0210 hdisk6
0010 0210 hdisk7

0011 0211 hdisk6
0012 0211 hdisk7
The LV now has only one copy on the new set of PVs, hdisk6 and hdisk7. To confirm that no
PPs are still used on the old set of LUNs, verify the USED PPs attribute. These attributes can
be view in the lspv command output (Example 8-114).
Example 8-114 Used PPs validation

# lspv hdisk2
PHYSICAL VOLUME: hdisk2 VOLUME GROUP: itsovg
PV IDENTIFIER: 0004afca0b8dedbd VG IDENTIFIER
0004afd80000d70000000131205383d4
PV STATE: active
STALE PARTITIONS: 0 ALLOCATABLE: yes
PP SIZE: 32 megabyte(s) LOGICAL VOLUMES: 0
TOTAL PPs: 511 (16352 megabytes) VG DESCRIPTORS: 1
FREE PPs: 511 (16352 megabytes) HOT SPARE: no
USED PPs: 0 (0 megabytes) MAX REQUEST: 256 kilobytes
FREE DISTRIBUTION: 103..102..102..102..102
USED DISTRIBUTION: 00..00..00..00..00
MIRROR POOL: None
Removing old volumes from the VG

After you verify that no PPs are still used on the old set of PVs, those volumes can be
excluded from the VG.
To remove the old volumes, use the C-SPOC commands to maintain consistency across all
cluster nodes. The removal is done using the smitty interface as shown in Example 8-115.
Example 8-115 Reducing itsovg VG using C-SPOC smitty interface “smitty cl_reducevg”
+--------------------------------------------------------------------------+
| Select the Volume Group that will hold the new Logical Volume |
| |
| |
| |
+--------------------------------------------------------------------------+
+--------------------------------------------------------------------------+
| |
| |

| |
+--------------------------------------------------------------------------+
Remove a Volume from a Volume Group

[Entry Fields]
VOLUME names hdisk2
FORCE deallocation of all partitions on no +
this physical volume?

On the second smitty panel, select a single PV for exclusion. The same process needs to be
repeated for every volume of the old set. Exclusion of the volume must be done only on one
node. The update is then propagated by C-SPOC to all nodes in the cluster.
To verify that all the volumes from the old set are excluded from the VG, run the lspv
command on the cluster nodes (Example 8-116). Configuration on all the cluster nodes
should be consistent.
Example 8-116 Verifying consistency of PVs configuration of PowerHA cluster nodes

itso_engine1
# lspv
hdisk2 0004afca0b8dedbd None
hdisk6 0004afca24cd59c3 itsovg concurrent
hdisk7 0004afca24cd5ae2 itsovg concurrent
hdisk0 0004afca82d256bb old_rootvg
hdisk1 0004afca95554544 rootvg active
hdisk3 0004afca0b8e068a None
hdisk4 0004afca0b8e136c None
hdisk5 0004afca0b8e1f63 None
itso_engine2
# lspv
hdisk2 0004afca0b8dedbd None
hdisk6 0004afca24cd59c3 itsovg concurrent
hdisk7 0004afca24cd5ae2 itsovg concurrent
hdisk0 0004afd87db8511f old_rootvg
hdisk1 0004afd8955a5e21 rootvg active
hdisk3 0004afca0b8e068a None

hdisk4 0004afca0b8e136c None
hdisk5 0004afca0b8e1f63 None
All PVs from the old set of LUNs are now removed from the itsovg VG configuration.
Replacing communication devices in PowerHA heartbeat networks

If some of the volumes in the old LUN set were configured as communication devices (in disk
heartbeat networks), those devices must be removed before a corresponding hdisk device
can be removed. After the old communication devices are removed, new communication
devices must be configured on the new set of LUNs. If you try to remove old hdisks before
removing the corresponding communication devices, an error message indicates that the
specified device is busy and will not be removed. As shown in Example 8-117, the
/dev/hdisk2 device is configured as communication interface in the net_diskhb_01 network,
and /dev/hdisk3 in net_diskhb_02.
Example 8-117 PowerHA topology information

NODE itso_engine1:
NODE itso_engine2:
In the example, two heartbeat networks are configured. Update one heartbeat network at a
time so that one heartbeat network is always available.
To remove a communication device from a heartbeat network, use the C-SPOC interface as
shown in Example 8-118. The same result can be achieved using the smitty C-SPOC
interface.
Example 8-118 C-SPOC communication device removal using C-SPOC CLI interface
# /usr/es/sbin/cluster/utilities/clrmnode -a'itso_engine2_hdisk2_01'
WARNING: Serial network [net_diskhb_01] has 1 communication device(s) configured.
Two devices are required for a serial network.
# /usr/es/sbin/cluster/utilities/clrmnode -a'itso_engine1_hdisk2_01'
Network removed: net_diskhb_01

After one pair of communication devices and the corresponding heartbeat network are
removed, you can add another pair of communication devices, one per cluster node. This new
pair must be configured using hdisk devices on the new set of LUNs.
To configure the new communication devices and heartbeat network correctly, use the
C-SPOC smitty interface on all participating cluster nodes. The interface allows you to select
devices and validates consistency across cluster nodes. To get to the smitty panel that
defines the heartbeat communication devices, perform the following steps:
1. Start the interface using the smitty hacmp command.
2. Select Extended Configuration  Extended Topology Configuration  Configure
HACMP Communication Interfaces/Devices  Add Communication
Interfaces/Devices  Add Discovered Communication Interface and Devices 
Communication Devices.
3. Select a communication device candidate as shown in Example 8-119.
Example 8-119 C-SPOC smitty interface for creating disk heartbeat network
+--------------------------------------------------------------------------+
| Select Point-to-Point Pair of Discovered Communication Devices to Add |
| |
| |
| # Node Device Pvid |
| drs_engine1 hdisk2 0004afca346ba911 |
| drs_engine1 hdisk4 0004afca356ea2c6 |
| drs_engine2 hdisk4 0004afca356ea2c6 |
| > drs_engine1 hdisk6 0004afca346ba892 |
| > drs_engine2 hdisk6 0004afca346ba892 |
| |
+--------------------------------------------------------------------------+
Repeat this process for any other heartbeat networks. After the reconfiguration of
net_diskhb_02, the test cluster topology configuration looks as shown in Example 8-120.
Example 8-120 PowerHA cluster topology after replacing communication devices

NODE drs_engine1:
drs_engine1_hdisk6_01 /dev/hdisk6
drs_cluster_svc 9.153.1.226
drs_engine1_boot 9.153.1.227

NODE drs_engine2:
drs_cluster_svc 9.153.1.226
drs_engine2_boot 9.153.1.228
The communication device replacement is now complete for both heartbeat networks.
The last step is to propagate the updates from the node where the changes were made to all
other nodes in the cluster. This step can be done either trough the smitty interface (smitty
cm_ver_and_sync.select) or the equivalent CLI command
(/usr/es/sbin/cluster/utilities/cldare -rt -V 'normal'). After the verification is
completed, ensure that the topology configuration on the other nodes in the cluster is identical
to the node used for heartbeat network configuration changes.
Deleting the device definitions from the server ODM

To remove device definitions, use the C-SPOC interface to maintain consistency across all
cluster nodes. The removal is done using the smitty interface as shown in Example 8-121.
Example 8-121 Removing devices using C-SPOC interface “smitty cl_disk_man.rem.nodes”

+--------------------------------------------------------------------------+
| Select Node(s) to Remove Disk From |
| |
| |
| > itso_engine1 |
| > itso_engine2 |
| |
+--------------------------------------------------------------------------+
+--------------------------------------------------------------------------+
| Select A Disk To remove |
| |
| |
| 0004afca0b8dedbd ( hdisk2 on all cluster nodes ) |
| 0004afca0b8e068a ( hdisk3 on all cluster nodes ) |
| 0004afca0b8e136c ( hdisk4 on all cluster nodes ) |
| 0004afca0b8e1f63 ( hdisk5 on all cluster nodes ) |
| |
+--------------------------------------------------------------------------+

Remove a Disk From the Cluster

[Entry Fields]
Nodes drs_engine1,drs_engin>
Disk 0004afca0b8dedbd
KEEP definition in database no +

If you are not planning on reusing the devices, change the KEEP definition in database
attribute to no to remove the ODM device entries.

Verify that all disk definitions from the old LUN set are now removed from the cluster nodes
using lsdev -Cc disk and mpio_get_config -Av commands (Example 8-122).
Example 8-122 Results from issuing the lsdev -Cc disk and mpio_get_config -Av commands
# lsdev -Cc disk
hdisk0 Available 04-08-00-5,0 16 Bit LVD SCSI Disk Drive
hdisk1 Available 04-08-00-8,0 16 Bit LVD SCSI Disk Drive
hdisk6 Available 00-08-02 MPIO Other DS4K Array Disk
hdisk7 Available 00-08-02 MPIO Other DS4K Array Disk
Frame id 0:
Controller count: 2
Partition count: 1
Partition 0:
hdisk6 11 B (preferred) ITSO_vol_t001
hdisk7 12 A (preferred) ITSO_vol_t002

Unassigning old set of DS5000 LUNs from the cluster nodes
The DS5000 Storage Manager GUI window used for LUN mapping removal is shown in
Figure 8-29.
Figure 8-29 Unassigning DS5000 LUNs from cluster nodes
The LUN mapping removal must be repeated for all the LUNs from the old LUN set. After you
remove all the LUNs, the test migration is complete.
8.8.4 Migration Backout Considerations

The data migration process described in 8.8.3, “Detailed migration steps using PowerHA
C-SPOC LVM mirroring” on page 355 removes the LV copy. After the LV copy is removed, you
can no longer use that copy. The only accessible copy of the LV is on the new set of LUNs. If,
for whatever reason, you decide to go back to the old set of LUNs, you need to repeat the
migration process in the reverse direction. Depending on the amount of data, this process
might take a significant amount of time. The downside of this method is that at the end of the
migration, you have only one physical copy of the data. During the planning phase, before the
migration starts, you need to decide whether the risk associated with having only one copy of
data is acceptable.
The alternative approach allows you to preserve an extra copy of the data. The splitlvcopy
command allows you to split a copy from the original LV and creates an LV from it. To maintain
consistency of data, however, this command requires the application to be stopped and the
file system to be unmounted. If you decide that having an extra copy is more important than
doing the migration nondisruptively, use splitlvcopy.
The splitlvcopy command does not allow you to select which copy is used to form the new
LV. It automatically uses the last copy of the LV. In the example, the command uses the copy
on the new set of LUNs. If this is not the result you want, rename the original LV and then
rename the new LV to use the original LV name. This way, the file system uses the copy on
the new set of LUNs. If you need to go back to the original set of LUNs, all you need to do is
to swap the LV names again. This process does not require any data to be copied.

8.9 Data migration using Linux LVM2 mirroring
Unlike Logical Volume Management (LVM) migration on other operating systems, Linux LVM2
can create a mirrored logical volume, but cannot split the mirror after it is synchronized.
Instead, Linux LVM can move physical extents in use by a volume group and its logical
volumes to the new target devices using a block-by-block copy. This process uses an
underlying mirror volume, but it does so on a segment basis and not an entire volume.
Restriction: This method has been presented as an “online” method (where application
I/O can continue to run, while the migration is taking place) in some RedHat publications.
However, testing with heavy I/O loads has found that migration can take a long time or stall
altogether. Therefore, stop application I/O, or schedule the migration when access to the
source volumes is light.
8.9.1 Introduction to Linux LVM2

Linux LVM is architecturally similar to other volume management facilities on other operating
systems in terms of its storage virtualization and operating flexibility.
The Logical Volume Management on Linux is called LVM. The latest version of the LVM is
called LVM2 and is compatible with earlier versions of Linux LVM1 except for certain
clustering and snapshot features.
LVM2 requires the device mapper kernel driver, a generic device mapper that provides the
framework for the volume management. For more information about Linux LVM2, see:
http://tldp.org/HOWTO/LVM-HOWTO/
In this scenario, the system already has Linux LVM2 installed and the volumes to be migrated
are part of an existing LVM volume group.
Linux LVM can move physical extents used by a volume group and its logical volumes to the
new target devices using the pvmove command.
The pvmove command creates a temporary, mirrored logical volume and uses it to copy the
data from the source device to the target device in segments. The original logical volume
metadata is updated to use the temporary mirror segment until the source and target
segments are in sync. It then breaks the mirror and the logical volume uses the target location
for that segment. This process is repeated for each segment to be moved. After all segments
are mirrored, the temporary mirror logical volume is removed and the volume group metadata
is updated to use the new target volumes.
8.9.2 Multipathing considerations

The example uses the DS8000 Multipath Subsystem Device driver to provide multipathing
and load balancing for the DS8000 LUNs on the host system. There is another alternative for
multipathing with Linux called DM-Multipath or DM-MPIO. For more information about
DM-MPIO and for a comparison between SDD and DM-MPIO, see “Considerations and
Comparisons between IBM SDD for Linux and DM-MPIO” available at:
http://www-1.ibm.com/support/docview.wss?uid=ssg1S7001664

8.9.3 Linux LVM migration scenario
The following environment and scenario are used to illustrate a Linux LVM migration.
General outline of the Linux LVM migration

The following are the high-level steps used to migrate data with Linux LVM. The source
volumes are already in an LVM volume group.
1. Configure target storage volumes on the DS8000.
2. Configure the host to recognize the DS8000 volumes.
3. Install Multipath Subsystem Device Driver (SDD).
4. Modify LVM to recognize the SDD devices.
5. Partition the SDD vpath devices.
6. Configure lvm.conf to recognize SDD vpath devices.
7. Initialize SDD vpath devices to the LVM.
8. Extend the existing LVM volume group with the new vpath partitions.
9. Stop the application I/O and unmount the file systems. Use pvmove to migrate extents off
the old device to the new device in the volume group.
10.Remove the old source devices from LVM.
11.Remove the old source devices from the system. Removing the devices is a disruptive
process, and can be done at the next maintenance window.
Migration environment
The physical test environment is composed of the following components:
򐂰 Application Server: RedHat AS4, 2.6.9-42.ELsmp kernel
򐂰 Fibre Channel HBA: Qlogic qla2340 HBAs
򐂰 Source Storage Controller: EMC Symmetrix
򐂰 Source volumes:
– /dev/sdd1: A primary partition on a 34-GB Symmetrix device
– /dev/sde1: A primary partition on a 9-GB Symmetrix device
– /dev/sdf1: A primary partition on a 9-GB Symmetrix device
򐂰 Target Storage Controller: IBM System Storage DS8300
򐂰 Target Volumes: A 36 GB and two 9 GB volumes are configured on the DS8000 to be used
as targets for the migration.
򐂰 LVM version2
򐂰 LVM configuration:
– One volume group
– Two logical volumes on the volume group
– Two file systems
Important: Back up the data on your source volumes before you start the migration
process.
LVM configuration before migration

The current LVM configuration for the example consists of a single LVM volume group, called
migvg, with two physical volumes with a single primary partition, /dev/sdd1 and /dev/sde1.

Two logical volumes, /dev/migvg/medium_lv (34 G) and /dev/migvg/small_lv (8 GB), are
built on migvg (Example 8-123). Note the total number of extents and free extents available on
each physical volume. The extents will be migrated from the old physical volumes to the new
physical volumes in the volume group.
Example 8-123 LVM Configuration before migration

[root@x345-tic-4 /]# vgdisplay -v migvg
Using volume group(s) on command line
Finding volume group "migvg"
--- Volume group ---
VG Name migvg
System ID
Format lvm2
Metadata Areas 2
Metadata Sequence No 37
VG Access read/write
VG Status resizable
MAX LV 0
Cur LV 2
Open LV 0
Max PV 0
Cur PV 2
Act PV 2
VG Size 42.09 GB
PE Size 32.00 MB
Total PE 1347
Alloc PE / Size 1344 / 42.00 GB
Free PE / Size 3 / 96.00 MB
VG UUID edL624-ocV6-q128-QCT6-7kkC-34Q6-8twJsB
--- Logical volume ---

LV Name /dev/migvg/medium_lv
VG Name migvg
LV UUID uHwA8J-4f2Y-RsXZ-4rHi-inyl-e3G2-y3LV06
LV Write Access read/write
LV Status available
# open 0
LV Size 34.00 GB
Current LE 1088
Segments 2
Allocation inherit
Read ahead sectors 0
Block device 253:3

LV Name /dev/migvg/small_lv
VG Name migvg
LV UUID TxMb0I-Ou9V-kGRX-1Mlu-kASE-32km-v1zBs3
LV Status available
# open 0
LV Size 8.00 GB
Current LE 256
Segments 1
Allocation inherit

Block device 253:4

PV Name /dev/sdd1
PV UUID 54dWUW-OI1N-sHJC-wFTE-VCE2-xT4b-2sqZQx
PV Status allocatable
Total PE / Free PE 1078 / 0
PV Name /dev/sde1
PV UUID 99jSJO-DZi1-29hN-PwR3-ovFA-ga0N-0XDiAE
Configuring target storage on DS8000 using DSCLI

Physically cable the DS8000 to your SAN and zone each HBA port on your Linux host to the
assigned ports from the DS8000. For best results, use no more than one host initiator per
zone. However, more than one storage port can be a member of the zone. Ideally, you would
have multiple zones with one host initiator, and one or more storage initiators in each zone for
multiple logical paths. Multipath Subsystem Device Driver (SDD), which is installed later,
handles the I/O and load balancing on each logical path.
Interactive mode DSCLI is used to configure the storage in this example. For more
information about DSCLI, see DS8000 Command-Line Interface User’s Guide,
SC26-7625-05. To configure the target storage, perform the following steps:
1. Create an empty DS8000 Volume Group to hold the DS8000 LUNs (Example 8-124).
Example 8-124 Create an empty volume group on the DS8000

dscli> mkvolgrp -hosttype LinuxRHEL x345-tic-4-vg
Date/Time: March 6, 2007 11:43:20 AM CET IBM DSCLI Version: 5.2.400.426 DS: IBM.2107-75L4741
CMUC00030I mkvolgrp: Volume group V0 successfully created.
2. Create fixed block DS8000 volumes and assign them to the Volume Group
(Example 8-125).
Example 8-125 Create and verify volumes on the DS8000

dscli> mkfbvol -extpool p4 -cap 37 -type ds -name x345-tic-4_#h -volgrp v0 2000
Date/Time: March 6, 2007 12:16:31 PM CET IBM DSCLI Version: 5.2.400.426 DS: IBM.2107-75L4741
CMUC00025I mkfbvol: FB volume 2000 successfully created.
dscli> mkfbvol -extpool p4 -cap 10 -type ds -name x345-tic-4_#h -volgrp v0 2001
CMUC00025I mkfbvol: FB volume 2001 successfully created.
dscli> lsfbvol -volgrp v0
Name ID accstate datastate configstate deviceMTM datatype extpool cap (2^30B) cap (10^9B) cap (blocks)
===================================================================================================================
x345-tic-4_2000 2000 Online Normal Normal 2107-900 FB 512 P4 37.0 - 77594624
3. Create Host Definitions and assign the Volume Group to the host (Example 8-126).
Example 8-126 Create and verify the host on DS8000 using DSCLI
dscli> mkhostconnect -wwname 210000e08b879f35 -volgrp v0 -hosttype LinuxRHEL x345-tic-4-qla0

CMUC00012I mkhostconnect: Host connection 0000 successfully created.
dscli> mkhostconnect -wwname 210000e08b0f381a -volgrp v0 -hosttype LinuxRHEL x345-tic-4-qla0
CMUC00012I mkhostconnect: Host connection 0001 successfully created.
dscli> lshostconnect 0 1
Name ID WWPN HostType Profile portgrp volgrpID ESSIOport
=============================================================================================
x345-tic-4-qla0 0000 210000E08B879F35 LinuxRHEL Intel - Linux RHEL 0 V0 all
x345-tic-4-qla0 0001 210000E08B0F381A LinuxRHEL Intel - Linux RHEL 0 V0 all
Configuring the Linux host to recognize DS8000 volumes

Use procedures appropriate for your Linux Distribution and Host Bus Adapter Vendor to add
device definitions for the new storage devices from the host system. This example uses the
Qlogic dynamic reconfiguration utility to dynamically add the new DS8000 devices as shown
in Example 8-127.
Example 8-127 Dynamic discovery of new DS8000 devices

[root@x345-tic-4 ql-dynamic-tgt-lun-disc-2.2]# ./ql-dynamic-tgt-lun-disc.sh --scan
Please make sure there is no active I/O before running this script
Do you want to continue: (yes/no)? yes
Issuing LIP on host2
Scanning HOST: host2
......
......
Found
2:0:1:0
2:0:1:1
2:0:2:0
2:0:2:1
3:0:0:0
3:0:0:1
3:0:1:0
3:0:1:1
Installing System Storage Multipath Subsystem Device Driver (SDD)

SDD usually comes included with your DS8000 as part of the DS8000 Licensed Microcode
Bundle. You can also obtain System Storage Multipath Subsystem Device Driver (SDD) for
your host at:
http://www-01.ibm.com/support/docview.wss?rs=540&uid=ssg1S7001350
Install SDD on the Linux host using the Linux rpm command as shown in Example 8-128.
Example 8-128 Installing Multipath Subsystem Device Driver (SDD) on Linux

[root@x345-tic-4 tmp]# rpm -ivh IBMsdd-1.6.2.0-1.i686.rhel4.rpm
Preparing... ########################################### [100%]
1:IBMsdd ########################################### [100%]
Added following line to /etc/inittab:
srv:345:respawn:/opt/IBMsdd/bin/sddsrv > /dev/null 2>&1
[root@x345-tic-4 tmp]# sdd start

Starting IBMsdd driver load:
[ OK ]
Issuing killall sddsrv to trigger respawn...
Starting IBMsdd configuration: [ OK ]
List the vpath devices and their underlying sd devices using the lsvpcfg command
(Example 8-129).
Example 8-129 lsvpcfg output

[root@x345-tic-4 tmp]# lsvpcfg
000 vpatha ( 252, 0) 75L47412000 = 6005076305ffc7860000000000002000 = /dev/sdg /dev/sdi /dev/sdk /dev/sdm
001 vpathb ( 252, 64) 75L47412001 = 6005076305ffc7860000000000002001 = /dev/sdh /dev/sdj /dev/sdl /dev/sdn
For more information about SDD, see Multipath Subsystem Device Driver User's Guide,
SC30-4131-01.
Partitioning SDD vpath devices

Partition your SDD vpath devices with your preferred partitioning utility. Remember that your
target vpath partitions must be large enough to accommodate the extents from the source
volumes. Set your partition type to 8e, Linux LVM.
In the example, fdisk was used to create a primary partition on each vpath and to set the
partition type (Example 8-130).
Example 8-130 Target vpath partitions

[root@x345-tic-4 /]# fdisk -l /dev/vpatha
Disk /dev/vpatha: 39.7 GB, 39728447488 bytes

64 heads, 32 sectors/track, 37888 cylinders
Units = cylinders of 2048 * 512 = 1048576 bytes
Device Boot Start End Blocks Id System

/dev/vpatha1 1 37888 38797296 8e Linux LVM
[root@x345-tic-4 /]# fdisk -l /dev/vpathb
Disk /dev/vpathb: 10.7 GB, 10737418240 bytes

64 heads, 32 sectors/track, 10240 cylinders
Units = cylinders of 2048 * 512 = 1048576 bytes
Device Boot Start End Blocks Id System

/dev/vpathb1 1 10240 10485744 8e Linux LVM
Modifying LVM configuration file to recognize SDD devices

The /etc/lvm/lvm.conf file is the configuration file for LVM2. It is used during LVM
initialization and LVM to recognize devices. To modify the /etc/lvm/lvm.conf file to recognize
the SDD devices, perform the following steps:
1. Modify the filter section to accept the vpath devices, but not the underlying sd devices that
make up the path (Example 8-131 on page 382). In the example, vpath devices, the sd
devices that belong to the root logical volume, and the migration source devices are
accepted. All other devices are rejected.

Remember: The lvm.conf filter section is read sequentially where “a” means accepted
and “r” means rejected (ignored). The filters are regular expressions. Devices that do not
match a pattern in the filter are automatically accepted.
Make sure that your filter accepts the vpath devices and rejects the underlying DS8000 sd
disks. The LVM returns an error about duplicate physical volumes if both the vpath and
underlying sd disk are seen.
2. Add an entry for vpath in the types section. This entry adds vpaths as a block device type
recognized by the LVM (Example 8-131).
Example 8-131 Filter and type settings in the /etc/lvm/lvm.conf file

filter = [ "a/vpath*/", "a/sda1/", "a/sda2/", "a/sdb1/", "a/sdd1/", "a/sde1/", "r/.*/" ]
...
types = [ "vpath", 16 ]
For more information about the /etc/lvm/lvm.conf file, see the Linux man page on
lvm.conf.
3. Run vgscan to allow the LVM to recognize the new disks (Example 8-132).
Example 8-132 vgscan

[root@x345-tic-4 /]# vgscan
Reading all physical volumes. This may take a while...
Found volume group "migvg" using metadata type lvm2
Found volume group "VolGroup00" using metadata type lvm2
Initializing the SDD vpath devices to the LVM

Initializing the SDD vpath devices prepares the disk for use by the LVM as a physical volume
(PV). The pvcreate command initializes the device, in the example a disk partition, as shown
in Example 8-133.
Example 8-133 Initialize vpath partitions for LVM

[root@x345-tic-4 /]# pvcreate /dev/vpatha1
Physical volume "/dev/vpatha1" successfully created
[root@x345-tic-4 /]# pvcreate /dev/vpathb1
Physical volume "/dev/vpathb1" successfully created
You can use a whole disk as an LVM physical volume. However, the LVM metadata written at
the beginning of the disk will not be recognized by outside operating systems. It is therefore at
risk of being overwritten.
The initialization process fails on a disk with an existing partition table on it. Double-check that
you are using the correct disk for the LVM if you see this error.
Tip: You can use the following commands to delete the partition table on the disk if you
want to use it for LVM. Use these commands with caution, as they will block access to the
existing data on the disk.
# dd if=/dev/zero of=/dev/diskname bs=1k count=1
# blockdev --rereadpt /dev/diskname

Extending the existing LVM volume group with the new vpath partitions
The new vpath devices need to be added to the existing LVM volume group, migvg. Add the
devices with the vgextend command as shown in Example 8-134.
Example 8-134 Using vgextend to add vpath devices to the volume group
[root@x345-tic-4 /]# vgextend migvg /dev/vpatha1
Volume group "migvg" successfully extended
[root@x345-tic-4 /]# vgextend migvg /dev/vpathb1
Volume group "migvg" successfully extended
Use the vgdisplay command to verify the addition of the new vpath devices (Example 8-135).
Notice the free Physical Extents (PE) available to the migvg volume group on each new
device.
Example 8-135 pvdisplay of the new vpath devices

[root@x345-tic-4 /]# pvdisplay /dev/vpatha1
--- Physical volume ---
PV Name /dev/vpatha1
VG Name migvg
PV Size 36.97 GB / not usable 0
Allocatable yes
PE Size (KByte) 32768
Total PE 1183
Free PE 1183
Allocated PE 0
PV UUID QKFc5R-tGRt-wX85-IJeJ-o0r3-8T62-DtHuo1
[root@x345-tic-4 /]# pvdisplay /dev/vpathb1

--- Physical volume ---
PV Name /dev/vpathb1
VG Name migvg
PV Size 9.97 GB / not usable 0
Allocatable yes
PE Size (KByte) 32768
Total PE 319
Free PE 319
Allocated PE 0
PV UUID AZweK9-jut4-T8fp-T9P8-7rLz-L2MQ-36958U
Using pvmove to migrate extents within the volume group

To use pvmove to migrate the extents within the VG, perform the following steps:
1. Stop all application I/O and sync your file systems. The file systems can remain mounted.
However, to maintain consistency, make no new writes.
2. Start the migration of extents using the pvmove command. It migrates extents within a
volume group from a source device to a target device as shown in Example 8-136 on
page 384. Issue the command:
pvmove source target
Where source and target are your source and target devices in the volume group.
Important: The pvmove command takes a long time to run because it performs a
block-by-block copy.

Example 8-136 pvmove in progress
[root@x345-tic-4 /]# pvmove /dev/sdd1 /dev/vpatha1
Detected pvmove in progress for /dev/sdd1
Ignoring remaining command line arguments
/dev/sdd1: Moved: 1.1%
..
If pvmove needs to be stopped during the migration, the pvmove --abort command can be
issued. The extents are spread between the old device and the new device. The migration
can be continued by issuing a pvmove command without any parameters.
3. After pvmove is complete, verify the migration with the vgdisplay command as shown in
Example 8-137. Notice that all the extents are free on /dev/sdd1 and are moved to
/dev/vpatha1.
Example 8-137 vgdisplay showing the extents migrated from device sdd1 to vpatha1
[root@x345-tic-4 ~]# vgdisplay -v migvg
VG Name migvg
System ID
Format lvm2
Metadata Areas 2
VG Status resizable
MAX LV 0
Cur LV 2
Open LV 2
Max PV 0
Cur PV 2
Act PV 2
VG Size 42.09 GB
PE Size 32.00 MB
Total PE 1347

VG Name migvg
LV UUID Uwat6I-N7eL-zwCK-kGlV-0lRO-yh5x-X7qzv0
LV Status available

# open 1
LV Size 34.00 GB
Current LE 1088
Segments 2
Allocation inherit
Block device 253:3

VG Name migvg
LV UUID AEk2Bg-E3Yp-R5x6-tbyt-8Njv-z9ve-kah7qI
LV Status available
# open 1
LV Size 8.00 GB
Current LE 256
Segments 1
Allocation inherit
Block device 253:4

PV Name /dev/sdd1
PV Name /dev/sde1
PV UUID QKFc5R-tGRt-wX85-IJeJ-o0r3-8T62-DtHuo1
4. Repeat the pvmove command to move extents from /dev/sde1 to /dev/vpathb1. Continue
with any additional devices with extents that need to be migrated.
5. Verify that the extents are now on the vpath devices using vgdisplay or lvdisplay
Example 8-138 vgdisplay command showing physical volumes with all extents moved to vpath
devices
PV Name /dev/sdd1

PV Name /dev/sde1
PV UUID tWVdi1-EpRZ-Nu5d-R27e-IhMq-AJgl-GqmSwa
Removing the old source device from LVM

After all the physical extents are migrated, the physical volume can be removed from the
volume group using the vgreduce command (Example 8-139).
Example 8-139 vgreduce to remove a physical volume from a volume group

[root@x345-tic-4 /]# vgreduce migvg /dev/sdd1
Removed "/dev/sdd1" from volume group "migvg"
[root@x345-tic-4 /]# vgreduce migvg /dev/sde1
Removed "/dev/sde1" from volume group "migvg"
Verify that the new vpath devices remain part of the volume group as shown in
Example 8-140.
Example 8-140 Verify migvg on vpath devices

[root@x345-tic-4 /]# vgdisplay -v migvg
VG Name migvg
System ID
Format lvm2
Metadata Areas 2
VG Status resizable
MAX LV 0
Cur LV 2
Open LV 2
Max PV 0
Cur PV 2
Act PV 2
VG Size 42.09 GB
PE Size 32.00 MB
Total PE 1347

VG Name migvg
LV UUID Uwat6I-N7eL-zwCK-kGlV-0lRO-yh5x-X7qzv0
LV Status available
# open 1
LV Size 34.00 GB
Current LE 1088
Segments 2
Allocation inherit
Block device 253:3

VG Name migvg
LV UUID AEk2Bg-E3Yp-R5x6-tbyt-8Njv-z9ve-kah7qI
LV Status available
# open 1
LV Size 8.00 GB
Current LE 256
Segments 1
Allocation inherit
Block device 253:4

PV UUID tWVdi1-EpRZ-Nu5d-R27e-IhMq-AJgl-GqmSwa
The logical volumes can now be remounted and application I/O started, or you can remove
the old devices from the system.
Removing the old source device from the system

This operation can be done at the next available maintenance downtime because application
I/O needs to be stopped and file systems unmounted.
1. Stop application I/O and unmount vpath devices.
2. Stop SDD. The underlying mapping between the DS8000 LUNs and the vpath devices will
not be lost (Example 8-141).
Example 8-141 Stop SDD

[root@x345-tic-4 /]# sdd stop
Stopping IBMsdd: [ OK ]

3. Change SAN zoning to remove access to old source volumes. This change preserves the
volumes on the old storage controller if they need to be restored.
4. Use procedures appropriate for your operating system and Host Bus Adapter Vendor to
remove device definitions for the old storage controller devices from the host system.
The example uses the Qlogic dynamic reconfiguration utility to dynamically remove the old
storage controller devices as shown in Example 8-142.
Example 8-142 Qlogic dynamic reconfiguration tool to remove old storage devices
[root@x345-tic-4 ql-dynamic-tgt-lun-disc-2.2]# ./ql-dynamic-tgt-lun-disc.sh -r -s
Please make sure there is no active I/O before running this script
Do you want to continue: (yes/no)? yes
.............
.............
Removed
2:0:0:0
2:0:0:14
2:0:0:23
2:0:0:3
5. Start SDD (Example 8-143).
Example 8-143 Start SDD

[root@x345-tic-4 ~]# sdd start
Starting IBMsdd driver load:
[ OK ]
Issuing killall sddsrv to trigger respawn...
Starting IBMsdd configuration: [ OK ]
6. Remount LVM logical volumes.

7. Verify SDD configuration using datapath commands as shown in Example 8-144.
Example 8-144 Datapath query device command output

[root@x345-tic-4 ~]# datapath query device
Total Devices : 3
DEV#: 0 DEVICE NAME: vpatha TYPE: 2107900 POLICY: Optimized Sequential

SERIAL: 75L47412000
============================================================================
0 Host2Channel0/sdg OPEN NORMAL 1 0
1 Host2Channel0/sdp OPEN NORMAL 0 0
2 Host3Channel0/sdi OPEN NORMAL 40 0
3 Host3Channel0/sdl OPEN NORMAL 0 0
DEV#: 1 DEVICE NAME: vpathb TYPE: 2107900 POLICY: Optimized Sequential

SERIAL: 75L47412001
============================================================================

0 Host2Channel0/sdh OPEN NORMAL 39 0
1 Host2Channel0/sdq OPEN NORMAL 0 0
2 Host3Channel0/sdj OPEN NORMAL 0 0
3 Host3Channel0/sdm OPEN NORMAL 0 0
DEV#: 2 DEVICE NAME: vpathc TYPE: 2107900 POLICY: Optimized Sequential

SERIAL: 75L47412002
============================================================================
0 Host2Channel0/sdo OPEN NORMAL 1 0
1 Host2Channel0/sdr OPEN NORMAL 0 0
2 Host3Channel0/sdk OPEN NORMAL 38 0
3 Host3Channel0/sdn OPEN NORMAL 0 0
8. Restart application I/O.
8.10 Data migration using network block devices

Network block devices are block devices where the real device is attached to a different Linux
host. The device accepts normal block commands to write, read, and maintain data. However,
the driver forwards the I/O through a TCP network to the server that the actual device is
connected to. For migrations, the network block device can be used as transport mechanism
and then discarded when the migration is done.
Using this technique, it is possible to migrate data, block by block, to another destination. The
easiest way to migrate data to the new device is using shell copy commands like cp, dd, or
cpio. A more elegant way is to establish a mirror using the network devices and allow the
background synchronization copy the data to the new target.
The following example involves NDB and DRDB, which are both integrated in the Linux
kernel. Other implementations are also referenced.
8.10.1 General architecture

The Linux Network Block Device nbd, is a basic implementation of a generic network block
device that is included in the kernel since version 2.1.101. It provides read-only and
read/write access to a remote device. In read-only mode, many clients can mount the remote
device. For write access, only one host is allowed to access the remote device at a time.
There is no access control included in the code. Therefore, the data on the network device
can be deleted if more than one client tries to write to this device.

Figure 8-30 shows the architecture in which the migration using nbd will work. At the target
server, a nbd server is started for the device /dev/sdb1. The source server accesses this
device using the nbd-client. The new device is named /dev/nbd0.
The migration itself is operated at the source location.
Source Target
Location Location
nbd-client nbd-server
/dev/sda1 /dev/nbd0 /dev/sdb1
Figure 8-30 Migration architecture for network block devices
When the nbd server is started, it listens to a TCP or UDP port assigned by the administrator.
The server is started with the command shown in Example 8-145.
Example 8-145 Starting the nbd-server

# nbd-server 5000 /dev/sdb1
# _
In this example, the nbd device is represented by a file in the directory /export. When the
server starts, it listens at the port 5000 for incoming requests.
After the server is started, the client can create a block device by referencing the IP address
of the server and the port where the server listens (Example 8-146).
Example 8-146 Create a nbd device

# nbd-client 9.155.33.67 5000 /dev/nbd0
# _
The network block device is now ready to be used. For example, create a file system, mount it
and allocate files and directories. For data migrations, however, other methods are of
interests as shown in the following chapters.

8.10.2 Using shell commands
When the network block device is created, a file system can be allocated. Data can be
migrated, file by file, with cp or cp -R shell commands. You can also copy the whole device in
a single dd statement to the network device as shown in Example 8-147.
Example 8-147 Copying a raw device to the nbd

# dd if=/dev/sda2 of=/dev/nbd0 bs=1024
# _
This method requires that the applications using the data on the disk /dev/sda2 must be shut
down. In addition, the file system must be unmounted to catch the recent I/O from the cache.
Depending on the used capacity of the disk and the bandwidth of the network, this process
might take a while before all data is transmitted.
8.10.3 Using software RAID1

The easiest way of replicating the data using network devices is to establish RAID1 as a
software RAID using the tool mdadm. The mdadm tool can provide different RAID levels by using
real block devices. For data migration, RAID1 is mirroring between two volumes. After the
RAID1 is created, both volumes are synchronized in the background. When the
synchronization is completed, all updates by applications or users are replicated to both
devices. When the migration is finalized, close the applications, unmount the file systems, and
close the RAID1 using the mdadm tool.
To use software RAID1 to migrate data, perform the following steps:

1. Establish the RAID1 device (Example 8-148).
Example 8-148 Create the RAID1 out of the local disk and the nbd0 device
# mdadm --create /dev/md0 --level=mirror --raid-devices=2 /dev/sda2 /dev/nbd0
mdadm: /dev/sda2 appears to contain an ext2fs file system
size=1048576K mtime=Thu Jul 7 14:30:34 2011
Continue creating array? y
mdadm: array /dev/md0 started.
#
2. Close all applications and mount the file systems.

3. A new device is allocated that writes the data to both devices. Mount the file system to this
device.
4. Transmit the device /dev/sda2 to the network block device /dev/nbd0. The result is a new
device, /dev/md0, which is now the RAID1 device. The synchronization of both mirrors
starts immediately with this command. The progress of the synchronization can be
monitored with the command shown in Example 8-149.
Example 8-149 Query the current status of the synchronization

# cat /proc/mdstat
Personalities : [raid1]
md0 : active raid1 nbd0[1] dm-3[0]
1048512 blocks [2/2] [UU]
[======>..............] resync = 31.2% (327680/1048512) finish=0.5min
speed=23405K/sec

unused devices: <none>
# _
5. You can see a progress bar which shows the status of the synchronization. When the
synchronization is done, you see the status shown in Example 8-150.
Example 8-150 Status, when synchronization is done

# cat /proc/mdstat
1048512 blocks [2/2] [UU]

# _
6. The file system for the applications can be mounted at any time after the RAID1 was
established (Example 8-151).
Example 8-151 Mount the RAID1 device

# mount /dev/md0 /mnt
# df
Filesystem 1K-blocks Used Available Use% Mounted on
/dev/sda1 14761044 8922208 5089004 64% /
none 4071508 412 4071096 1% /dev
none 4076096 1932 4074164 1% /dev/shm
none 4076096 188 4075908 1% /var/run
none 4076096 0 4076096 0% /var/lock
none 4076096 0 4076096 0% /lib/init/rw
/dev/md0 1032088 103468 876192 11% /mnt <===
# _
7. Any updates from users or the applications are now mirrored to both underlaying devices.
When you are ready the cut over, close the applications and unmount the file systems,
then close the RAID1 as shown in Example 8-152.
Example 8-152 Cut over to the target device

# umount /mnt
#
# mdadm --stop /dev/md0
mdadm: stopped /dev/md0
#
# cat /proc/mdstat
# _
8. When the RAID1 was established, a md-superblock was written to each device. To mount
a remote device on the target system, this superblock must be removed using the
command shown in Example 8-153.
Example 8-153 Cleanup the superblock at the target system

# mdadm --zero-superblock /dev/nbd0
#

# mount /dev/sda1 /APPLICATION
# _
The migration is now complete. The remaining operations are to stop the nbd-server at the
target system, stop the nbd-client at the source system, and clean up the RAID1.
8.10.4 Summary
The Network Block Device (nbd) can be used for a basic migration to new Linux systems and
new storage devices. For migrations involving only a few disks, this type of migration can easy
to implement.
However, two major issues need to be considered first:

򐂰 The communication between the server and the client is realized by specifying a TCP user
port. This communication can be a security issue in some environments.
򐂰 There is no control mechanism that prevents that more than one client from write data to
the server. The lack of a control mechanism is a high risk to corrupt the data on the server.
You need to make sure that no other client has access to the server device.
8.10.5 Using Linux LVM2 mirroring

Another method to migrate data to a network block device is using the Linux Logical Volume
Manager LVM. The source data devices must already be part of an existing LVM
configuration. Otherwise, the devices itself must first be migrated to LVM.
The volume group is then extended by the new nbd device. Transmit the data using the
methods as described in 8.9, “Data migration using Linux LVM2 mirroring” on page 376.

9
Chapter 9. Using TDMF for z/OS

This chapter describes the TDMF for z/OS product. It addresses the steps required to use the
product in a typical local migration scenario.

򐂰 TDMF for z/OS overview
򐂰 Customer scenarios
򐂰 TDMF preferred practices

9.1 TDMF for z/OS overview
Transparent Data Migration Facility (TDMF) for z/OS is a host-based software solution for
data migration in a z/OS environment. TDMF offers the following functions:
򐂰 Is user-initiated and controlled
򐂰 Allows for full system sharing throughout the data center
򐂰 Guarantees full access to the data at any point during a migration operation
򐂰 Supports dynamic takeover on the part of the target device
9.1.1 Data migration and migration tool definitions and characteristics

Working with TDMF for z/OS includes the following characteristics:
򐂰 Multiple volume migrations can be established during a single session.
򐂰 The migration tool is dynamically activated and terminated.
򐂰 Applications remain unaware that migration is underway. The data on the source device is
continuously and fully accessible for read and write activity.
򐂰 After the synchronization of source and target volumes is complete, the takeover of the
target device is nondisruptive.
򐂰 The tool supports a multiple-system shared data environment.
򐂰 The tool guarantees complete physical data integrity.
򐂰 The use of the tool is not restricted to any control unit model type or device type. Except as
noted, all devices that are online in the data center can participate in a migration session
as required.
򐂰 All volumes (source and target) of a migration must be online.
򐂰 No data set can be allocated to a target volume during migration.
Restriction: TDMF for z/OS does not allow access to the target volume during the
migration process.
򐂰 A source volume might not contain an active local page data set or swap data set.
򐂰 The source and target volumes must be of the same track geometry.
These characteristics represent the ideal of a transparent and non-disruptive migration facility.
9.1.2 Terminology
These terms describe the TDMF installation and migration process:
򐂰 Master system: The TDMF system, running as an z/OS batch job or started task that is
responsible for the data copy function. There is only one master system in a TDMF
session.
򐂰 Agent system: An associated TDMF z/OS image running in a shared storage environment
with the master. To ensure data integrity, any z/OS LPAR with access to the volumes must
be run on either the master or an agent system. The master and associated agent
systems communicate through a shared system communications data set (COMMDS).
򐂰 Source volume: The DASD volume containing the data to be migrated.
򐂰 Target volume: The DASD volume receiving the migrated data.

򐂰 Migration pair: The source and target volumes for a single migration.
򐂰 Migration group: A group of volume pairs using the same group name.
򐂰 Migration session: A master system plus any agents, if applicable, and pairs or groups to
be processed in a single TDMF execution.
򐂰 Synchronization: The period when the following steps occur:
a. I/O is inhibited to the source devices (quiesce phase).
b. The last of the updates are collected and reflected to the target devices
(synchronization phase).
c. The source and target volume serial numbers are exchanged (redirect phase).
d. The I/O is resumed to the new source (original target) device (resume phase)
򐂰 Swap migration: A session in which the source and target VOLSERs are swapped and I/O
redirection to the new source occurs at volume termination. The original source device is
marked offline.
9.1.3 TDMF z/OS components architecture

Figure 9-1 depicts the TDMF z/OS architecture and components. Volumes can be moved
between different storage vendors or inside the same storage unit.
TDMF Architecture
Agent 3
Master SYSD
SYSA Agent 4 TDMF
Agent 2 SYSE
SYSC
Agent 1 TDMF TDMF
SYSB Agent 5
SYSF
ISPF ESCON/FICON
Monitor Director
Source Target
Source
Target
Source
Target
Communication
data sets
IBM ESS 2105 any vendor IBM DS8000
Figure 9-1 TDMF architecture (example with 6 LPARs)
The architecture has the following components:

򐂰 TDMF TSO monitor: The TDMF TSO monitor is used to view active or past migrations. It is
installed at the same time as the TDMF tool. This feature consists of REXX execs, which
require ISPF version 3.6 and TSO/E Version 2.4 at minimum.
Chapter 9. Using TDMF for z/OS 397

򐂰 Communications data set: The communications data set or COMMDS is used to pass
information between systems participating in a TDMF session. This data set contains the
status and messages related to a specific session. The COMMDS also serves as the input
file to the TDMF TSO monitor.
򐂰 Master and agents: Each LPAR that has access to the storage subsystem containing the
source and the target volumes must run an agent. Each master or agent within a session
must have access to the same COMMDS. Multiple agent systems can be involved in a
session locally.
Important: Because of a possible data integrity exposure, all systems accessing migration
volumes must be identified to the master system. TDMF includes various controls and
checks so you do not make the following errors:
򐂰 Assign or direct conflicting migrations to the same devices
򐂰 Attempt migrations to nonexistent devices
򐂰 Use the same COMMDS for two simultaneous or overlapping migration sessions
For audit purpose, do not use the same COMMDS for different sessions.
Master system responsibilities

The TDMF Master system is responsible for the following tasks:
򐂰 Initialize the TDMF Master system environment and the COMMDS
򐂰 Start and control each session for all participating systems
򐂰 Monitor source volume user I/O activity to detect updates
򐂰 Monitor target volume user I/O activity to prevent updates
򐂰 Copy data from the source volume to the target volume
򐂰 Process detected source volume updates from all systems
򐂰 Perform refresh operations to the target volume to reflect the update activity on the source
volume
򐂰 Check the internal health of the master environment and the health of all agent systems
Agent system responsibilities

The TDMF Agent system is responsible for the following tasks:
򐂰 Initialize the TDMF Agent environment and establish communications to the master
system through the COMMDS
򐂰 Acknowledge and process migration requests (volumes to be moved) from the master
system
򐂰 Monitor source volume user I/O activity and detect updates
򐂰 Monitor target volume user I/O activity to prevent updates
򐂰 Notify the master system of source volume update activity through the COMMDS
򐂰 Check the internal health of the agent environments and the health of the master system
9.1.4 TDMF z/OS process flow

The TDMF z/OS process is a phased approach to make sure that data integrity is maintained
in each phase. Data migration sessions can be interrupted at any time with no data loss.
During the migration progress, all updates are written to the original source volume and are
replicated by TDMF to the target volume. If an I/O violation occurs, the session is terminated.

TDMF system initialization process
Only after successful initialization of all systems in a TDMF session does migration proceed.
If any violation occurs during system initialization on any system defined in the session, no
migrations are performed.
If all systems in the session are not started within a 15-minute interval, the session does not
complete system initialization. If you start a system that is not part of an active session, TDMF
terminates the master job and all agent jobs using the same COMMDS.
If you use System Authorization Facility (SAF), and any volume involved in the migration
session fails SAF, the migration session itself fails.
SAF requirements are:

򐂰 Swap type migrations require ALTER authority on the source and target volumes.
򐂰 Point-in-Time migrations require READ authority for the source volume and UPDATE
authority for the target volume.
Volumes in a session can be terminated using the TDMF TSO Monitor or Batch Monitor on
the Master system before initialization of the agent systems. If you select the History option to
automatically record information about the migration session, the recording requires UPDATE
authority for the data set. For more information, see the System Authorization Facility (SAF)
and TDMF Installation and Reference, TDM-Z53IR.PDF at:
The master system initiates and controls all migrations. The agents must each phase to
proceed. If any system detects a violation, that specific migration terminates. Depending on
the state of the current migration, you must perform back-out processing to establish the
original status before the migration session started.

TDMF migration phases
The TDMF migration process flow can be broken into major phases as shown in Figure 9-2.
Initialization
I/O error
Activation
I/O error Copy
I/O error Refresh
I/O error
Quiesce
I/O error
Synchronize I/O error

I/O error
Redirect
Backout
Resume
Termination
Figure 9-2 Phases of a TDMF session
򐂰 Initialization phase: During this phase, all participating systems confirm the validity of the
source and target volumes. The phrase includes the following activities:
– Volume acknowledgement: Volumes that require an acknowledgement are not eligible
for confirmation and selection until one is received. This acknowledgement can use the
TDMF TSO Monitor, a Batch Monitor, or the IBM MVS™ Write-to-Operator/
Write-to-Operator with Reply (WTO/WTOR).
This acknowledgement is required for volumes being migrated whose disaster recovery
or mirror techniques are not compatible unless the
ALLOWmirrorchange(NOACKnowledge) option is specified. TDMF recognizes
volumes that use Peer-to-Peer Remote Copy (PPRC), extended remote copy (XRC),
TrueCopy from Hitachi, and Symmetrix Remote Data Facility (SRDF). These volumes
are recognized with or without Consistency Groups from EMC.
The disaster recovery or mirroring type must be compatible between the source and
target volumes. Compatibility includes the vendor command interface structure and
parameters. If they are not compatible, the ALLOWmirrorchange option must be
specified and an acknowledgement received.
– Volume confirmation: Volumes that require confirmation are not eligible for volume or
group selection until a confirmation is received. This confirmation can use the TDMF
TSO Monitor, a Batch Monitor, or the MVS Write-to-Operator/Write-to-Operator with
Reply (WTO/WTOR). The order of confirmation determines the order of volume
selection. Volumes or groups that do not require confirmation are immediately available
for volume or group selection.

– Volume selection: By default all volume pairs defined in a session are automatically
selected during the initialization phase. However, volume selection is affected when
certain user options are specified. The options that affect volume selection are volume
confirmation, number of concurrent volumes, active in copy, group options, and allow
mirroring changes.
– Volume initialization: Initialization of all volume level control blocks and page fixing of all
real storage frames are necessary for a volume migration.
򐂰 Activation phase: In this phase, the I/O monitor is activated by TDMF for the source and
target device. All systems in the session then attempt to allocate each source and target
volume, preventing them from being inadvertently varied offline.
򐂰 Copy phase: In this phase, the master system begins a COPY volume task to copy data
from the source volume to the target volume. Each source volume has an independent
COPY volume task in the migration session.
If there are source volume updates during the copy phase, the master system collects the
updated information to be processed during the refresh phase. Changes to the source
volume are kept in a bitmap located in the COMMDS.
򐂰 Refresh phase: When the COPY volume phase completes one pass of the source volume,
the target volume receives the updates made to the source volume. This phase is
controlled by the master system.
Multiple refresh phases occur until TDMF determines that synchronization of the target
volume can be achieved, at which time the master system quiesces the source volume.
򐂰 Quiesce phase: In this phase, the master system stops all I/O activity to the source
volume. When an agent system receives a quiesce request, the agent system sends the
master system the final group of detected updates.
򐂰 Synchronize phase: When all the systems comply with the quiesce request, all detected
updates are written to the target volume.
򐂰 Volume I/O redirect phase: In this phase, TDMF redirects I/O activity from the source
volume to the target device. The master system requests that all systems do a redirect,
and confirms that these redirects are successful. When this process is complete, the
master system rewrites the volume labels to change the serial numbers and performs
redirect processing.
Tip: If an LPAR in the same TDMF session failed an I/O redirection, TDMF resets the
redirection on all LPARs. TDMF goes on to the resume phase with an I/O error.
򐂰 Resume phase: Immediately after successful I/O redirect processing, the master system
performs RESUME processing and initiates the RESUME request for the agent systems.
Resuming allows user I/O to continue to the volume on its new device. After all systems
process the RESUME request, the original (source) device is marked offline.
If an I/O error during the quiesce, synchronize, or I/O redirection phases, the source
device is re-enabled for I/O and the UCB quiesce is removed.
򐂰 Terminate phase: When a volume completes a migration, the fixed storage of that volume
is released for possible reuse in the current session. All dynamic allocations are also
removed.
9.1.5 TDMF z/OS hardware compatibility

TDMF is designed to support any Count Key Data/Extended (CKD/E) capable control units as
a source or a target storage subsystem. These functions include Hyper-Volumes,

Flexible-Volumes, and all user-defined volume sizes. Handling of Extended Address Volumes
(EAVs) depends on the z/OS level. See the TDMF Reference Manual for more details.
Because TDMF is a host software migration tool, there are no hardware prerequisites.
9.1.6 Installing TDMF z/OS

There are two ways to install TDMF on a system:
򐂰 With SMP/E for z/OS
򐂰 Without SMP/E (preferred for migration service)
Pre-Installation considerations
The master system requires additional storage, depending on the number of volumes in the
session. This additional storage is not a significant amount, and is only allocated for volumes
with the Parallel Access Volumes (PAV) option set:
򐂰 10 or less volumes = 4 K
򐂰 10 - 32 volumes = 4 K to 12 K
򐂰 32 - 64 volumes = 12 K to 24 K
򐂰 64 - 128 volumes = 24 K to 48 K
򐂰 128 - 256 volumes = 48 K to 96 K
򐂰 256 - 512 volumes = 96 K to 192 K
The Softek TDMF for z/OS Installation Assurance Guide for V5R3.0, TDM-Z53IA-001.pdf,
has complete pre-installation and post-installation checklists that need to be followed. Be sure
to read the copy of this document that matches the level of software you are installing. It is
available at:
http://www-950.ibm.com/services/dms/en/support/tdmf/zos/v5/data/TDM-Z53IA-001.pdf
Installing with SMP/E

An installation standard procedure must be provided to indicate to SMP/E where to locate all
the required data sets. Unless predefined within the consolidated software inventory (CSI)
using DDDEFs, allocate all the DDnames for the Softek TDMF target and distribution
libraries.
The following members in the distributed SAMPLIB can be copied and tailored to install
Softek TDMF. The DLIBZONE and TARGZONE must be updated in the samples to reflect the
zone definitions for the site.
The order of installation is:

1. ALLCSMP creates the SMP/E files LOG, LOGA, MTS, PTS, SCDS, and STS.
2. INITCSI creates and initialize the CSI, global, target, and distribution zones.
3. DDDEF creates the DDDEF entries in SMP/E.
4. SMPE is the sample procedure that the following jobs run.
5. SMPEREC performs an SMP/E RECEIVE of the Softek TDMF product.
6. SMPEAPK performs an SMP/E APPLY CHECK of the Softek TDMF product.
7. SMPEAPP performs an SMP/E APPLY of the Softek TDMF product.
8. SMPEACK performs an SMP/E ACCEPT CHECK of the Softek TDMF product.
9. SMPEACC performs an SMP/E ACCEPT of the Softek TDMF product.
The following procedure creates a complete and separate SMP/E environment for Softek
TDMF. Alternatively, you can install the product in any other SMP/E structure, However, you
must edit the jobs to fit your environment.

To install Softek TDMF using SMP/E, perform the following steps:
1. Edit ALLCSMP with the TDMFEDIT exec and submit (allocates MTS, PTS, SCDS, STS,
LOG, and LOGA data sets).
2. Edit SMPE with the TDMFEDIT exec and copy to PROCLIB.
Note: In JES3 environments, you might need to separate this step into multiple jobs.
3. Edit INITCSI with the TDMFEDIT exec and submit (calls SMPE).
4. Edit DDDEF with the TDMFEDIT exec and submit.
5. Edit SMPEREC with the TDMFEDIT exec and submit.
6. Edit SMPEAPK with the TDMFEDIT exec and submit.
7. Edit SMPEAPP with the TDMFEDIT exec and submit.
8. Edit SMPEACK with the TDMFEDIT exec and submit.
9. Edit SMPEACC with the TDMFEDIT exec and submit.
Installing without SMP/E

For data migration purposes, it is easier to install without SMP/E. In this case, the INSTALL
member in the sample library performs an IEBCOPY upload of the modules. TDMF
maintenance is not provided, and you must therefore install the full package again. For more
information, see TDMF Installation and Reference (Maintenance Overview).
Note: If the installation has a security package on the z/OS system on which TDMF is
installed, you must modify the security package. Otherwise TDMF will not run properly.
9.1.7 TDMF z/OS customer requirements

TDMF supports all z/OS-based operating systems that are currently supported by IBM. This
product is not suitable for use with IBM z/VM® and IBM VM/ESA®, VM/XA, VSE/ESA,
VSE/XA, or DOS. The Operating Systems Support Matrix is at the following web address:
http://www-950.ibm.com/services/dms/en/support/tdmf/zos/matrix.html
If a z/OS system is available, copy z/VM, Linux for System z, and IBM z/VSE® volumes
(3390-x volumes only) using REPLICATE instead of MIGRATE. These operating systems
must be OFFLINE because there is no AGENT available. Define the source and target
volumes to the z/OS system and run the same jobs as in z/OS, but with parameter
REPLICATE instead of MIGRATE. After the data is replicated, use ICKDSF REFORMAT to
rename the target volumes to their original VOLIDs.
Special considerations must be taken when z/OS is running under z/VM when allocating the
COMMDS. For more information, see the TDMF installation manual at:

Tip: Perform periodic checks of the Required IBM Maintenance, IBM Technical Support
(PTFs), and Technical Information Bulletins (TIBs). These requirements must be
implemented to ensure successful TDMF operation.
Required IBM Maintenance:

http://www-950.ibm.com/services/dms/en/support/tdmf/zos/apars.pdf
IBM Technical Support:

http://www-950.ibm.com/services/dms/en/support/tdmf/zos/v5/ptf530.html
Technical Information Bulletins:

http://www-950.ibm.com/services/dms/en/support/tdmf/zos/v5/tibs530.html
Security
If the installation has a security package on the z/OS system on which TDMF is installed, you
need to modify the security package. Otherwise TDMF does not run properly. Check the
profiles and command tables.
Limiting access to the TDMF authorized library to prevent unauthorized use of the TDMF
system can be accomplished through security packages. The user for the SYSOPTN batch
job (provided by the TDMF installation) must have UPDATE authority for the library indicated
by the SECCOM DD statement. The master migration job also needs UPDATE authority for
the SECCOM file if the authorization key specifies volume or terabyte limits. To update
authorization keys using Option 10 of the TDMF TSO Monitor as shown in Figure 9-3, you
must have UPDATE authority for the TDMLLIB library. In addition, the library must be
allocated as SECCOM in the SYSOPTN job.
Softek TDMF Enterprise Row 1 to 1 of 1

Command ===> Scroll ===>
Company : IBM DEUTSCHLAND GMBH

Site . : IBM MAINZ
Site ID : 11660
Enterprise Number . . . . . . . . . . : 50071
Enterprise Effective Date . . . . . . : January 10, 2011
Enterprise Expiration Date . . . . . . . December 31, 2011
Return Code from Security . . . . . . : 00000000
Features Enabled . . . . . . . . . . . : MIGRATE REPLICATE
PPIT OVA TCP/IP
Commit Changes? (yes/no) . . . . . . . . ___

Softek TDMF for z/OS Version 5.3.0 Base Level.
No Security Key Key Description Added Date/Time By
01 E01785ABE1402405 ENTERPRISE # 50071 01/10/2011 20:59 SYSOPTN

Figure 9-3 TDMF license keys
If the History option is selected, UPDATE authority is required for the data set specified.
When viewing the history file (and any COMMDS) using the TDMF TSO Monitor, you must
have READ authority.

System Authorization Facility
If you want to use the System Authorization Facility (SAF), select VOLUME SECURITY =
YES when using the SYSOPTN batch job in SAMPLIB. For more information about the TDMF
system defaults, see TDMF System Defaults in the TDMF Installation Manual at:
For a swap migration, ALTER authority must be in effect for the source and target volumes.
Error messages are issued for all volumes not meeting these requirements in a session.
TDMF checks for two types of classes:

򐂰 CLASS=DATASET for the COMMDS history data set and the TDMF load library defined
on the SECCOM DD statement
򐂰 CLASS=DASDVOL for volumes allowed in a pairing
If the authorization keys specify volume or terabyte limits, the TDMLLIB (SECCOM DD
statement) must have update authority for the user ID submitting the jobs.
9.2 Customer scenarios

This section illustrates a simple example of migrating data from an ESS 800 to a DS8000. In
the example four volumes are moved, but the process would be the same for any number of
volumes.

The test environment has the following characteristics:
򐂰 Hardware and software setup as shown in Figure 9-4
Example: Customer scenario

TDMF
TSO
Monitor
Master
CECE
Agent 1 TDMF
BVS2
Communication
Dataset Migrate
Source
Dev: 6000-6003
VOLID:
RS6000 Target
RS6001 Dev: 8100-8103
RS6002 VOLID:
RS6003 XX8100
XX8101
XX8102
XX8103
IBM ESS 800

IBM DS8100
Figure 9-4 Example: Graphical overview (Customer scenario)

򐂰 z/OS level 01.12.00 (Figure 9-5).
IEE421I RO *ALL,D IPLINFO 955

MCECEBC RESPONSES --------------------------------------
IEE254I 18.27.15 IPLINFO DISPLAY 954
SYSTEM IPLED AT 12.41.01 ON 01/21/2011
RELEASE z/OS 01.12.00 LICENSE = z/OS
USED LOADB9 IN SYS1.IPLPARM ON CA13
ARCHLVL = 2 MTLSHARE = N
IEASYM LIST = BC
IEASYS LIST = 00
IODF DEVICE: ORIGINAL(CA13) CURRENT(CA13)
IPL DEVICE: ORIGINAL(CA10) CURRENT(CA10) VOLUME(CEBCR1)
MZBCVS2 RESPONSES --------------------------------------
IEE254I 18.27.15 IPLINFO DISPLAY 855
SYSTEM IPLED AT 13.35.54 ON 01/27/2011
RELEASE z/OS 01.12.00 LICENSE = z/OS
USED LOADS9 IN SYS1.IPLPARM ON CA13
ARCHLVL = 2 MTLSHARE = N
IEASYM LIST = S2
IEASYS LIST = 00
IODF DEVICE: ORIGINAL(CA13) CURRENT(CA13)
IPL DEVICE: ORIGINAL(CA10) CURRENT(CA10) VOLUME(CEBCR1)
Figure 9-5 Display IPL information including the z/OS level

򐂰 Two LPARs (LPAR 4 and 5) as seen in Figure 9-6
IEE421I RO *ALL,D M=CPU 973

MCECEBC RESPONSES ----------------------------------------
IEE174I 18.31.15 DISPLAY M 972
PROCESSOR STATUS
ID CPU SERIAL
00 + 04423A2064
01 + 14423A2064
02 + 24423A2064
03 + 34423A2064
04 + 44423A2064
05 + 54423A2064
06 + 64423A2064
CPC ND = 002064.116.IBM.51.00000002423A
CPC SI = 2064.116.IBM.51.000000000002423A
CPC ID = 00
CPC NAME = P002423A
LP NAME = BVS1 LP ID = 4
MIF ID = 4
MZBCVS2 RESPONSES -------------------------

IEE174I 18.31.15 DISPLAY M 857
PROCESSOR STATUS
ID CPU SERIAL
00 + 05423A2064
01 + 15423A2064
02 + 25423A2064
03 + 35423A2064
04 + 45423A2064
05 + 55423A2064
06 + 65423A2064
CPC ND = 002064.116.IBM.51.00000002423A
CPC SI = 2064.116.IBM.51.000000000002423A
CPC ID = 00
CPC NAME = P002423A
LP NAME = BVS2 LP ID = 5
MIF ID = 5
Figure 9-6 Displaying processor information
򐂰 DISK subsystems: IBM TotalStorage ESS 800 and IBM System Storage DS8000
򐂰 Channel attachment (ESCON / FICON) for storage

9.2.2 Migration steps
To migrate data using TDMF for z/OS, perform the following steps:
1. Verify that all volumes to be migrated are online to all LPARs involved in the migration as
shown in Figure 9-7. Send a display command for the source devices to all LPARs to
query the status and volser. In the example, one volume is offline on one of the LPARs.
RO *ALL,D U,,,6000,4
IEE421I RO *ALL,D U,,,6000,4 112
MCECEBC RESPONSES ---------------------------
IEE457I 22.56.27 UNIT STATUS 111
UNIT TYPE STATUS VOLSER VOLSTATE
6000 3390 A RS6000 PRIV/RSDNT
6001 3390 O RS6001 PRIV/RSDNT
6002 3390 A RS6002 PRIV/RSDNT
MZBCVS2 RESPONSES ---------------------------
6003 3390 OFFLINE RS6003 PRIV/RSDNT
Figure 9-7 Display source device address at all LPARs
2. Bring the volume online by issuing a VARY ONLINE command to all LPARs for all devices
in the range (6100-6103) as shown in Figure 9-8.
RO *ALL,VARY 6000-6003,ONLINE
IEE421I RO *ALL,VARY 6000-6003,ONLINE 163
MCECEBC RESPONSES ---------------------------------------------------
MZBCVS2 RESPONSES ---------------------------------------------------
Figure 9-8 Check source device address

3. After verifying all the source devices, check the target devices by issuing the same display
command to all LPARs for the target range of devices (8100-8103). Figure 9-9 shows that
all the devices are offline.
IEE421I RO *ALL,D U,,,8100,4 216

8100 3390 F-NRD /RSDNT
Figure 9-9 Check target device address
4. Bring the target devices online by issuing the VARY ONLINE command to the devices as
seen in Figure 9-10.
IEE421I RO *ALL,VARY 8100-8103,O 240

8100 3390 O XX8100 PRIV/RSDNT
8100 3390 O XXS8100 PRIV/RSDNT
Figure 9-10 Vary target devices online

5. After all the devices (source and target) used in the data migration are online, start the
TDMF TSO monitor. The member to start the monitor is in the TDMF library
(SYS1.IBM.HGTD530.GTDELIB) data set. Open the data set in edit to run the REXX member
as shown in Figure 9-11.
DSLIST - Data Sets Matching SYS1.IBM.HGTD530.* Row 1 of 7

Command ===> Scroll ===> CSR
Command - Enter "/" to select action Message Volume

------------------------------------------------------------------------------
e SYS1.IBM.HGTD530.GTDELIB VSLNSM
SYS1.IBM.HGTD530.GTDLLIB VSLNSM
SYS1.IBM.HGTD530.GTDMLIB VSLNSM
SYS1.IBM.HGTD530.GTDPLIB VSLNSM
SYS1.IBM.HGTD530.GTDTLIB VSLNSM
SYS1.IBM.HGTD530.SAMPLIB VSLNSM
SYS1.IBM.HGTD530.TIB VSLNSM
***************************** End of Data Set list ***************************
Figure 9-11 Edit the TDMF exec file
6. Run the TDMF member to start the TDMF monitor as shown in Figure 9-12.
EDIT SYS1.IBM.HGTD530.GTDELIB Row 00001 of 00003

Name Prompt Size Created Changed ID
_________ GZDRECON
ex_______ TDMF 87 2011/06/30 2011/06/30 01:33:23 STRUBEL
_________ TDMFMON
**End**
Figure 9-12 Start the TDMF monitor

7. The TDMF startup window is displayed as seen in Figure 9-13. Press Enter to continue
the process.
Command ===>
TTTTTTTTTT DDDDDDDDD MM MM FFFFFFFFFF

TTTTTTTTTT DDDDDDDDDD MMM MMM FFFFFFFFFF
TT DD DD MMMM MMMM FF
TT DD DD MM MMMM MM FF
TT DD DD MM MM MM FFFFFFFF
TT DD DD MM MM FFFFFFFF
TT DD DD MM MM FF
TT DD DD MM MM FF
TT DDDDDDDDDD MM MM FF
TT DDDDDDDDD MM MM FF
Transparent Data Migration Facility for z OS Version 5.3.0

LICENSED MATERIALS - PROPERTY OF IBM
Press Enter to continue

Figure 9-13 TDMF startup window
8. The TDMF Monitor menu is displayed as seen in Figure 9-14. There are many options
presented in this menu. Each option is covered in detail in the TDMF Installation and
Reference manual. Select option 0 to open SYS1.IBM.HGTD530.SAMPLIB, where you find
example migration jobs.
Softek TDMF Monitor Menu

Option ===>
0 Change or Submit Replication or Migration Jobs

1 Current Sessions: Monitor Progress
2 Current Sessions: User Interaction and Status
3 Current Sessions: Display Messages
4 Current Sessions: Display Offline Volume Access (OVA)
5 Current/Past Sessions: Display Performance Data
6 Past Sessions: Display Summary
7 Past Sessions: Display Details
8 Past Sessions: Display Communication Dataset History
9 Display Installation Options and Environment
10 Display/Modify Installation Security Environment
11 Display/Modify User's TSO Monitor Options
12 Build Replication or Migration Jobs
U Support Utilities
H Help and Message Detail Facility
Figure 9-14 TDMF Monitor Menu

9. The migration jobs are in the CNTL data set (specifically STXXXXX.TDMF.ITSO.TDMF.CNTL in
the test environment). Edit (customize) some of the members. An example of editing the
command data set that defines the new COMMDS is shown in Figure 9-15.
EDIT STXXXXX.TDMF.ITSO.TDMF.CNTL(CDSL$G01) - 01.00 Columns 00001 00072

****** ***************************** Top of Data *****************************
000001 //CDSL$G01 JOB A185,'IBM',MSGCLASS=I,REGION=0M,
000002 // CLASS=A,TIME=1440,NOTIFY=&SYSUID
000003 //* telefon number
000004 //* IBM name of service people- TDMF DATA MIGRATION
000005 /*JOBPARM S=CEBC
000006 //******************************************************
000007 //* Create SYSCOM DATA SETS AT LOCAL SITE
000008 //******************************************************
000009 //STEP01 EXEC PGM=IEFBR14
000010 //SYSPRINT DD SYSOUT=*
000011 //SYSCOM DD DSN=SYS1.TDMF.ITSO.LOCAL.SYSCOM01,
000012 // DISP=(NEW,CATLG,DELETE),
000013 // SPACE=(CYL,70,,CONTIG),UNIT=3390,
000014 // DCB=(LRECL=4096,BLKSIZE=4096,RECFM=F,DSORG=PS),
000015 // VOL=SER=RS6100
000016 /*
****** **************************** Bottom of Data ***************************
Figure 9-15 Define new COMMDS

Figure 9-16 shows examples of a MASTER and an AGENT job. For more information
about parameters, see “Creating jobs” on page 426.
EDIT STXXXXX.TDMF.ITSO.TDMF.CNTL(TDMLBC01) - 01.01 Columns 00001 00072

****** ***************************** Top of Data *****************************
000001 //TDMLBC01 JOB A185,'IBM',MSGCLASS=I,REGION=0M,
000003 //* IBM telefon number
000005 /*JOBPARM S=CEBC
000006 //******************************************************
000007 //* Master JOB TDMF running: LOCAL
000008 //******************************************************
000009 //LOCAL01 EXEC PGM=GTDMAIN,PARM=MASTER,REGION=0M
000010 //STEPLIB DD DISP=SHR,DSN=SYS1.IBM.HGTD530.GTDLLIB
000011 //GTDKEY DD DISP=SHR,DSN=SYS1.IBM.HGTD530.GTDLLIB
000012 //SYSCOM DD DISP=SHR,DSN=SYS1.TDMF.ITSO.LOCAL.SYSCOM01
000014 //SYSSNAP DD SYSOUT=*
000015 //SYSIN DD *
000016 SESSION M(CEBC) AGENT(CVS2)
000017 OPT(FAST UNIDENT(T) NOPROMPT CONC(2)
000018 PAC CHECKT)
000019 MIGRATE RS6000 XX8100 YY6000
000020 MIGRATE RS6001 XX8101 YY6001
000021 MIGRATE RS6002 XX8102 Yy6002
000022 MIGRATE RS6003 XX8103 YY6003
****** **************************** Bottom of Data ***************************
****** ***************************** Top of Data *****************************

000001 //TDALCV01 JOB A185,'IBM',MSGCLASS=I,REGION=0M,
000003 //* IBM telefon number
000005 /*JOBPARM S=CVS2
000006 //******************************************************
000007 //* AGENT JOB TDMF running: CVS2
000008 //******************************************************
000009 //STEP01 EXEC PGM=GTDMAIN,PARM=AGENT,REGION=0M
000010 //STEPLIB DD DISP=SHR,DSN=SYS1.IBM.HGTD530.GTDLLIB
000011 //GTDKEY DD DISP=SHR,DSN=SYS1.IBM.HGTD530.GTDLLIB
000012 //SYSCOM DD DISP=SHR,DSN=SYS1.TDMF.ITSO.LOCAL.SYSCOM01
000014 //SYSUDUMP DD SYSOUT=*
****** **************************** Bottom of Data ***************************
****************************************************************
** Start AGENT, on for each LPAR connected to source and target
****************************************************************
Figure 9-16 Example: submit Master and Agent jobs

10.Before starting, run the job to create the COMMDS. Figure 9-17 shows a new query of the
data sets that displays the COMMDS (SYS1.TDMF.ITSO.LOCAL.SYSCOM01).
DSLIST - Data Sets Matching SYS1.TDMF.ITSO Row 1 of 1


------------------------------------------------------------------------------
SYS1.TDMF.ITSO.LOCAL.SYSCOM01 RS6100
Figure 9-17 Checking for new COMMDS
11.After verifying that the new COMMDS exists, submit the master and agent jobs as shown
in Figure 9-16 on page 413.
12.Change to the TDMF TSO monitor using option 1 in the Monitor menu as seen in
Figure 9-14 on page 411. Option 1 allows you to monitor the progress of the current
session (Figure 9-18). In this example, the data is being migrated on four volumes. Two of
the volumes are copying at the same time. The progress monitor is updated to show the
migration phases and the percentage of completion for each phase after pressing the
Enter key. You also get an estimated end time for the initial copy phase for each volume.
Session Monitor Row 1 to 8 of 8

Command ===> Scroll ===> PAGE
Softek TDMF Master V5.3.0 Session Active.

ComDataSet : SYS1.TDMF.ITSO.LOCAL.SYSCOM01
Source Migration Percent Complete --------->
Group VolSer Phase ..10...20...30...40...50...60...70...80...90..100
RS6000 Inactive
RS6001 Inactive
RS6002 Inactive
RS6003 Inactive

RS6000 Copy ---------------> 0:01:04
RS6001 Copy ---------------> 0:01:04
RS6002 Inactive
RS6003 Inactive
Figure 9-18 Monitor the session

Figure 9-19 displays the output of the session monitor with the first two volumes in a
Refresh phase at 100% complete. The copy has not yet started for the next two volumes.


RS6000 Refresh 1 ------------------------------------------------>
RS6001 Refresh 1 ------------------------------------------------>
RS6002 Inactive
RS6003 Inactive
Figure 9-19 Refresh phase started
Figure 9-20 shows that the first two volumes are completed, and the next two volumes are
in the copy phase. Based on the parameter (CONCurant=2), only two volumes run at a
time. However, this parameter can be dynamically changed in the monitor.


RS6000 Complete
RS6001 Complete
RS6002 Copy --------------> 0:01:04
RS6003 Copy ---------------> 0:01:03
Figure 9-20 Complete and copy phases

Figure 9-21 Illustrates the Refresh to Synchronize phase for the second set of volumes.


RS6000 Complete
RS6001 Complete
RS6002 Refresh 1 ------------------------------------------------>
RS6003 Refresh 1 ------------------------------------------------>
- - - - - - - - - - - - - -
RS6000 Complete
RS6001 Complete
RS6002 Quiescing ------------------------------------------------>
RS6003 Quiescing ------------------------------------------------>
- - - - - - - - - - - - - -
RS6000 Complete
RS6001 Complete
RS6002 Synchronize ------------------------------------------------>
RS6003 Synchronize ------------------------------------------------>
Figure 9-21 Refresh - Synchronize phase
Figure 9-22 shows that all the copies for this session are complete. Remember that this
migration was a simple one involving only the four volumes. A typical migration scenario
usually involves many more volumes.
Session Monitor ** no sessions active **

No Softek TDMF sessions active.

Figure 9-22 TDMF session complete

13.Check the state of the volumes using the display command again. Figure 9-23 lists the
output in the syslog (only the user syslog in these examples). As you can see, the old
source volumes 6100-6103 are now offline to the LPARs. The volume ID is displayed even
if the device is offline, which is expected at the completion of the TDMF data migration.
RO *ALL,D U,,,6000,4
IEE421I RO *ALL,D U,,,6000,4 978
6000 3390 OFFLINE YY6000 /RSDNT
******************************** BOTTOM OF DATA ********************************
Figure 9-23 Query old source volumes after migration
Figure 9-24 shows the output of a query against the old target volumes (8100-8103).
These volumes are now the source volumes and online to the LPARs. The data migration
is complete.
IEE421I RO *ALL,D U,,,8100,4 007

MCECEBC RESPONSES --------------------------------------------
MZBCVS2 RESPONSES --------------------------------------------
******************************** BOTTOM OF DATA ***************
Figure 9-24 Verify that the new source volumes are online

9.3 TDMF preferred practices
This section offers general advice and preferred practices for TDMF z/OS.
9.3.1 Keeping current

Make sure before any migration that your TDMF z/OS code is up-to-date. Use the following
techniques to keep your code current:
򐂰 Bookmark the TDMF Technical Support website. It contains information that can help you
successfully run TDMF.
򐂰 Register for automatic email notification. Registration ensures that you are notified
whenever the Technical Support website is updated. You receive the actual link for
registration with your license key (by mail).
򐂰 Click Notification Registry on the Technical Support website to get an IBM Lotus®
Notes® Group ID.
򐂰 Download each program temporary fix (PTF) as it becomes available. If you miss one,
however, the list fixes are still available. All PTFs are available for a SMP/E installation
only. NON SMP/E installation requires a new download of the code and reinstallation. See
the Installation and Reference Manual at:
򐂰 Manuals are periodically updated. Make sure that you have the most current information
available by checking for Technical Information Bulletins (TIBs). TIBs are typically issued
between manual releases. They are available at:
9.3.2 Setting default options

As part of the post-installation tailoring of TDMF, option defaults are set by the SYSOPTN
batch job and the loading of the software keys. If you want to use pacing in every migration
session, set this value to YES in the SYSTOPTN batch job during first activation. That value is
the default for all subsequent sessions.
Values set in the SYSOPTN batch job can be overridden within a session. The order of
overriding of options is as follows:
򐂰 Session statement
򐂰 Group statements
򐂰 Migrate statements
In other words, values in the session statement override the installation defaults. Values in the
group statement override the session and installation defaults. And finally, the migrate
statements override all of the above.
For more information, see the TDMF Installation Manual, Chapter 2 “TDMF Control
Statements” at:

9.3.3 Storage requirements
In this example, there are 16 3390-3 volumes being migrated with two systems involved, and
Offline Volume Access is not selected. In this case, the storage requirements for the master
and agent systems are:
򐂰 Master system:
– Fixed common storage: 612K ECSA, 9K ESQA, 256-bytes CSA
– Pageable common storage: 28k ECSA
– Fixed extended private area storage: 16,600K
򐂰 Agent system:
– Fixed common storage: 617K ECSA, 5K ESQA, 256-bytes CSA
– Pageable common storage: 28k ECSA
– Fixed extended private area storage: 656K
If the Compare option or Full Speed Copy is requested, an additional 900K buffer for each
volume migration is allocated in ECSA. For more information about storage requirements, see
the TDMF Installation Manual, Chapter 2 “Storage Requirements” at:
9.3.4 Communications data set

Each TDMF session requires a unique communications data set (COMMDS). The master and
agent systems communicate to each other using the COMMDS. The COMMDS cannot be on
the same volume involved in the session. Allocate disk space for COMMDS on a separate,
NONSMS-managed volume. The COMMDS must be placed on a device that supports
CKD/E.
Important: If another session is submitted using the same COMMDS name while the first
session is running, unpredictable results occur.
If a problem occurs, the COMMDS is the primary tool for problem determination and
resolution. Although reuse of a COMMDS is normal, be aware that the previous session data
is no longer available. If there is a problem, do not reuse the COMMDS so it is available for
problem determination and resolution.
Tip: Use a new COMMDS for each session, so that if a problem occurs, all information is
available for analysis.
Member ALLOCCM in SAMPLIB allocates the COMMDS. This data set must be physically
located on a cylinder boundary with contiguous space. For an example, see Figure 9-15 on
page 412. Volumes containing COMMDS cannot be moved by TDMF
COMMDS reserve
TDMF periodically issues a RESERVE macro for the COMMDS to serialize communication
between the master and agent systems. For details, see the TDMF Installation Manual,
Chapter 4 “Unicenter CA-MIM Resource Sharing or Global Resource Serialization.”
Sizing a COMMDS
To calculate the size of a COMMDS, use the following formula:
COMMDS CYLS = V * (S + K)

Where V is the number associated with the number of volumes as follows:
򐂰 64 volumes = 2.5
򐂰 128 volumes = 5.0
򐂰 256 volumes = 7.5
򐂰 512 volumes = 10.0
In addition, S is the number of participating systems, and K is the size of the source volumes
involved as follows:
򐂰 For 3390-3, K = 4
򐂰 For 3390-9, K = 6
򐂰 For 3390-27, K = 15
For example, consider a system which contains 128 3390-3 and 128 3390-9 volumes across
8 LPARs. Using the largest device type in the session (and therefore K = 6), the size is
calculated as follows:
CYLS = 7.5 * (8 + 6) (always use the largest device type in session)
CYLS = 7.5 * 14
CYLS = 105 (round down if required)
Data set type and location for COMMDS

The COMMDS can be defined as a generation data group (GDG), which can ease the
tracking of these data sets. In this case, each new generation must be created before running
the session. Using GDG allows reference within the JCL to relative generation zero (0) and
the base name to be specified on the SESSION control statement.
Tip: GDG data sets are not preferred because they can be overwritten during the migration
process. Instead, use sequential data sets, which are the default.
For more details, see the TDMF Installation Manual, Chapter 3 “Placement of the
Communications data set.”
9.3.5 Participation of agent systems

In a shared DASD environment, all Logical Partitions (LPARs) must have access to the source
and target devices participating in the TDMF session. Each master or agent system is only
aware of I/O to the source or target device from its perspective. If a session is run without
inclusion of all LPARs connected to the source or target devices, data integrity is
compromised.
Since TDMF version 5.2.0, you can run an AGENT while the source and target device are
offline to the LPAR. Therefore, you do not have to split the volumes into groups where the
volumes have the same status to the LPARs. Some volumes (source and target) can be
offline in one LPAR and the session will not terminate. For more information, see the TDMF
Installation Manual.
The Unidentified Systems option can be used to verify that the correct number of agent
systems are being run. The default action is set when the SYSOPTN job is run, but can be
overridden using the options within the SESSION control statement. The options available
are:
򐂰 Terminate on Error: This issues an error message and terminates the migration/replication
(RC08).
򐂰 Warning: This issues a warning message, but the migration/replication continues (RC04).

򐂰 Informational: This issues an informational message and the migration/replication
continues (RC00).
This parameter works with only 3990-6 control units or later (2105/2107). If you are using
3990-3 (old HW), TDMF cannot check the attached systems.
Important: You are responsible for running the correct number of AGENTs.
For more details, see the TDMF Installation Manual, Chapter 2 “Session Options.” In addition,
a Technical Information Bulletin (TIB) is available at the following web address:
9.3.6 Protection of target volume data

If a target volume contains data, TDMF overwrites that data as long as no files are open at the
time the session starts. If a file is open on a target volume at volume initiation, TDMF
terminates that migration session. The system displays return code 12 (RC 12) and any error
messages pertinent to the situation.
The CHECK TARGET option ensures that TDMF does not overwrite data on a target device.
Selection of this option informs TDMF that only the VTOC, VTOCIX, and VVDS entries are
allowed on the target volume.
Tip: Use the CHECK TARGET option as default to protect your data. Set the CHECK
TARGET option in each session control statement.
For details, see the TDMF Installation Manual, Chapter 2 “Session Options.”
9.3.7 Pacing
Always use pacing so that application performance is not affected. Some volumes regularly
experience high I/O rates.
The following options are available to regulate pacing:

򐂰 Reverse pacing
򐂰 Full speed copy
򐂰 FastCopy
򐂰 Active in Copy
Pacing is active during the Copy and Refresh phases only. After the volumes or session enter
the Quiesce phase and continue through the Termination phase, pacing is not started. For
more details, see the TDMF Installation Manual, Chapter 1 “Major Phases of Migration.”
Tip: Generally, use the default pacing options. Move frequently updated volumes during a
low activity time period.
Reverse pacing
Reverse pacing starts at one track per I/O operation, and scales upwards depending on the
activity on the source volume. For more details, see the TDMF Installation Manual, Chapter 3
“Dynamic Volume Pacing.”

Full speed copy
This option causes TDMF to increase the number of buffers from one to two to allow
interleaving of the I/O. Using two buffers shortens the migration time. Use this option
judiciously because the additional load to the source and target volumes can negatively affect
applications and the migration itself. Field statistics show that the decrease in migration time
is 40% on average. For more details, see the TDMF Installation Manual, Chapter 2 “Full
Speed Copy Impact.”
FastCopy
FastCopy instructs TDMF to copy only the allocated tracks and cylinders, ignoring empty
ones. This option can decrease the amount of time required to migrate a volume with no
performance impact. Use the FastCopy option at all times. For more details, see the TDMF
Installation Manual, Chapter 2 “Common Options Table.”
Active in Copy
The volume pairs in a migration are considered active from volume initialization to volume
termination. Active in Copy instructs TDMF to treat a volume as being active from volume
initialization through the completion of the first refresh pass. Use of this option limits the
number of active copy tasks but speeds the normal copy process. For more details, see the
TDMF Installation Manual, Chapter 2 “Session Options.”
9.3.8 Rank contention and storage subsystem performance

When large numbers of volumes are to be copied, it is important to understand the following
about the system:
򐂰 The architecture of the subsystem
򐂰 The logical to physical mapping of the volumes within the array
Use this information during the planning, coding, and execution of TDMF sessions. This
information helps ensure that these sessions do not adversely affect application, array, and
subsystem performance.
When experimenting, do not migrate more than two volumes from a logical control unit (LCU).
Using monitors such as RMFMON, check if there is an impact to the system performance.
Also, check channel utilization to make sure that it is no more than 70-75% per channel path.
Change the number of concurrent running volumes based on the results using the TDMF
monitor.
For more details, see the TDMF Installation Manual, Chapter 3 “Raid Subsystems and Rank
Contention.”
9.3.9 TDMF interaction with other programs

Before the use or execution of TDMF, see the TDMF Installation Manual, Chapter 4: “Planning
Considerations.” Ensure that the conditions are in place for the coding and execution of the
TDMF sessions.
If you use an array-based mirroring solution in your environment, it must be protected.

Protection is especially important if you need to keep that environment active during a
migration. In this case, review the following topics in Chapter 4:
򐂰 Migrating from SRDF to PPRC Volumes
򐂰 Migrating to SRDF Volumes
򐂰 Peer-to-Peer Remote Copy

If GDPS PPRC (Hiperswap) is active, be aware that Hiperswap is switched off by TDMF at the
time of swap. See the Installation and Reference Manual for details.
9.3.10 Identification of volumes requiring special handling

Some volumes might require special handling before, during, or after a migration. These
volumes need to be identified and potentially put into one or more separate sessions.
In addition, some volume migrations require an application recycle to complete the migration.
For more information, see the TDMF Installation Manual, Chapter 4: “Planning
Considerations.”
Important: Do not move active Local Page data sets or active Sysplex Couple data sets.
Migrating Local Page data sets

Local Page data sets cannot be moved while active. To migrate them, perform the following
steps:
1. Analyze location of these data sets using the D ASM command at each LPAR accessing
the source/target storage (Example 9-1).
Example 9-1 Display Local Page data sets

COMMAND INPUT ===> /D ASM SCROLL ===> CSR
RESPONSE=MCECEBC
IEE200I 11.32.10 DISPLAY ASM 536
TYPE FULL STAT DEV DATASET NAME
PLPA 82% OK 5249 PAGE.MCECEBC.PLPA
COMMON 0% OK 5249 PAGE.MCECEBC.COMMON
LOCAL 0% OK 5249 PAGE.MCECEBC.LOCAL1
LOCAL 0% OK 524A PAGE.MCECEBC.LOCAL2
LOCAL 0% OK 5249 PAGE.MCECEBC.LOCAL3
LOCAL 0% OK 524A PAGE.MCECEBC.LOCAL4
PAGEDEL COMMAND IS NOT ACTIVE
2. Check whether the Local Page data sets are on separate volumes. If not, create
temporary Local Page data sets on different volumes (old storage subsystem).
3. Release all active Local Page data sets on the same volume using the PAGE DELETE
command. It takes time to release them from the system. Use the DISPLAY ASM
command to check on the progress.
4. When all Local Page data sets on the same volume are released from the system Page
Pool, migrate that volume.
5. After migrating the volume, use the PAGE ADD command to add Local Page data sets to
the System Page Pool again.
6. Repeat the proceeding steps for all volumes where Local Page data sets are located.
PLPA and COMMON Page data sets can be moved by TDMF, but must be moved separately.
Migrating JES spool and check point data sets

JES2 spool and checkpoint data sets cannot be moved with normal data. Mark all volumes
containing these data sets in your volume migration list and migrate them in the same
session. Run one volume at the time, and at a time when there is not a heavy I/O load on the

system. For more details on systems running JES3, see the TDMF Installation and Reference
Manual.
Migrating Couple data sets (sequence)

Analyze the volume where the Couple data sets are located using the command D XCF,
COUPLE,TYPE=SYSPLEX as shown in Example 9-2.
Example 9-2 Display sysplex Couple data sets

COMMAND INPUT ===> /D XCF,COUPLE,TYPE=SYSPLEX SCROLL ===> CSR
RESPONSE=MCECEBC
IXC358I 22.08.02 DISPLAY XCF 886
SYSPLEX COUPLE DATA SETS
PRIMARY DSN: SYS1.CEBCPLEX.XCF.CDS01
VOLSER: COUPL1 DEVN: B23D
FORMAT TOD MAXSYSTEM MAXGROUP(PEAK) MAXMEMBER(PEAK)
12/09/2010 10:43:02 32 200 (39) 303 (8)
ADDITIONAL INFORMATION:
ALL TYPES OF COUPLE DATA SETS ARE SUPPORTED
GRS STAR MODE IS SUPPORTED
CLUSTER RESOURCE MANAGEMENT IS SUPPORTED
SYSTEM STATUS DETECTION PROTOCOL IS SUPPORTED
ALTERNATE DSN: SYS1.CEBCPLEX.XCF.CDS02
VOLSER: COUPL2 DEVN: B23E
FORMAT TOD MAXSYSTEM MAXGROUP MAXMEMBER
12/09/2010 10:46:44 32 200 303
ADDITIONAL INFORMATION:
ALL TYPES OF COUPLE DATA SETS ARE SUPPORTED
GRS STAR MODE IS SUPPORTED
CLUSTER RESOURCE MANAGEMENT IS SUPPORTED
SYSTEM STATUS DETECTION PROTOCOL IS SUPPORTED
Check whether these data sets are on separate volumes. If not, do not move these volumes
because the time-out interval of the timer accessing the Couple Data Set (CDS) default is 15
seconds.
To check volumes also for Local Page data sets, perform the following steps:
1. Switch the couple data set from primary to alternate:
setxcf couple,type=sysplex,pswitch
2. Migrate the volume (COUPL1) with the non-active old primary couple data set.
3. Add the old primary to the system as the new alternate couple data set:
setxcf couple,type=sysplex,acouple=SYS1.CEBCPLEX.XCF.CDS01
4. Switch the couple data set again:
setxcf couple,type=sysplex,pswitch
5. Migrate the volume (COUPL2) with the non-active old alternate couple data set.
6. Add the old alternate to the system as the new alternate again:
setxcf couple,type=sysplex,acouple=SYS1.CEBCPLEX.XCF.CDS02

9.3.11 Migration considerations
Consider the following tips when planning your migration:
򐂰 Consider using a naming convention for the batch jobs. If a number of sessions are to be
run, use a naming convention that groups the master, agents, and COMMDS in an
easy-to-find format. For more information, see Chapter 11, “Using TDMF TCP/IP for z/OS”
on page 483.
򐂰 The keyword TIME=1440 or TIME=(mm,ss) can be specified on job cards to avoid system
ABEND 322 (S322). TIME=1439 is preferable because TIME=1440 disables SMF time
recording for that job. TIME=(mm,ss) allows recording of SMF times but limits the amount
of processor time used by the TDMF session. Ensure that a reasonable amount of time is
allowed for each TDMF session, especially if the COMPARE option is used. See members
MASTER and AGENT in the SAMPLIB for examples. For more information about the
TIME parameter, see the MVS JCL manuals in the z/OS Internet Library at:
http://www.ibm.com/servers/eserver/zseries/zos/
򐂰 Always code TDMF sessions for as many pairings as possible. The real storage
requirements, based on the number of concurrent volumes specified, remain the same
regardless of the number of volumes within the session. By coding the maximum number
of volumes per session, the management of multiple sessions is lessened.
򐂰 Set Active in Copy and the Number of Concurrent Volumes to the number of channels the
LPAR can access from the source subsystem. In a typical technology refresh, the target
subsystem has newer storage devices and newer channels, such as FICON versus
ESCON connections. Because ESCON channels are slower than FICON channels, limit
the number of active volumes based on the slower channel paths. A general rule of thumb
for ESCON channel paths is two active volumes per path.
򐂰 Employ FastCopy whenever possible. This option copies only the allocated tracks and
cylinders of the device: No empty tracks are copied.
򐂰 Use Full Speed Copy if you want to employ double buffering. When this option is selected,
two 900 K buffers are allocated and pacing is automatically started. Overall time reduction
is dependent on a number of factors, but an average of 45 percent can be expected.
򐂰 Select the Dynamic ICKDSF REFVTOC option. ICKDSF is dynamically called at the end
of the migration if invocation is necessary to reformat the VTOC. If no changes are
required, the call is not issued.
򐂰 Select Toleration of Invalid Count Fields to ensure that volumes with invalid count fields
(CCHH mismatch) are copied without error. These count fields are not corrected.

Creating jobs
Figure 9-25 shows creating a COMMDS.
//TDMFCX04 JOB (A185,SYS),'STXXXXX',

// NOTIFY=&SYSUID,REGION=0M,
// CLASS=A,MSGCLASS=X,MSGLEVEL=(1,1)
/*JOBPARM SYSAFF=CEBC
//****************************************************************
//** define new communication dataset
//** please set correct size in CYL
//** depending on number of volumes to move.
//** See TDMF Installation manual for details.
//****************************************************************
//DEFCOX04 EXEC PGM=IEFBR14
//DASDI1 DD DSN=SYS1.TDMF.ITSO.LOCAL.SYSCOM04,DISP=(NEW,CATLG),
// VOL=SER=RS6100,STORCLAS=NONSMS,
// UNIT=3390,SPACE=(CYL,70,,CONTIG)
/*
Figure 9-25 JCL example to create COMMDS
Sample job naming conventions are shown in Example 9-3. In a migration service, be aware
that you can run several jobs in parallel. A good idea is to reflect the type, LPAR name, and
the session number in the job name.
Example 9-3 Job Naming conventions

//TDMLBC04 JOB .....
!!! !
!!! +--> session number (2 digits)
!!+----> 2 char out of the LPAR name (first or last 2 char)
!! to identify AGENTs at different LPARs
!+-----> L=local
+------> M=Master, A=AGENT

Figure 9-26 shows creating a MASTER task.
//TDMLBC04 JOB (A204,SYS),'STXXXXX',

/*JOBPARM SYSAFF=CEBC
//** define master job
//****************************************************************
//MIGRAT04 EXEC PGM=GTDMAIN,PARM=MASTER,TIME=1440,REGION=0M
//STEPLIB DD DISP=SHR,DSN=SYS1.IBM.HGTD530.GTDLLIB
//SECCOM DD DISP=SHR,DSN=SYS1.IBM.HGTD530.GTDLLIB
//SYSCOM DD DISP=SHR,DSN=SYS1.TDMF.ITSO.LOCAL.SYSCOM04
//DSFPRINT DD SYSOUT=*
//SYSUDUMP DD SYSOUT=*
//SYSSNAP DD SYSOUT=*
//SYSIN DD *
SESSION
MASTER(CEBC)
AGENT(CVS2)
OPTIONS(CONCURRENT(02)
UNIDENTIFIEDSYSTEMS(TERMINATE)
CHECKTARGET
FASTCOPY
PACING(NORMAL)
ICKDSF
NOCONFIRM
NOPROMPT
NOALLOWINVALIDCOUNTS
NOCOMPARE
)
MIGRATE RS6000 XX8100 YY6000
/*
Figure 9-26 JCL example for master system

Figure 9-27 shows creating an AGENT task. The master and agent tasks together are called
a SESSION.
//TDALS204 JOB (A185,SYS),'STXXXXX',

/*JOBPARM SYSAFF=CVS2
//****************************************************************
//** Start AGENT, one for each LPAR connected to source and target
//****************************************************************
//AGENT04 EXEC PGM=GTDMAIN,PARM=AGENT
//STEPLIB DD DSN=SYS1.IBM.HGTD530.GTDLLIB,DISP=SHR
//SECCOM DD DSN=SYS1.IBM.HGTD530.GTDLLIB,DISP=SHR
//SYSCOM DD DSN=SYS1.TDMF.ITSO.LOCAL.SYSCOM04,DISP=SHR
//SYSSNAP DD SYSOUT=*
//SYSIN DD DUMMY
/*
Figure 9-27 JCL example for agent
Important: Do not forget that AGENTs must run on all LPARs. Also, make sure that the
volumes are accessed and online.
9.3.12 Estimating how long it takes to move the data

How long it takes to move the data depends on a number of factors. Some of the factors are
the options selected for a specific volume migration or for the entire session. Others are the
activity of the processor, channel paths, and devices. This section addresses the following
factors:
򐂰 Pacing
򐂰 Number of sessions
򐂰 Factors external to TDMF
򐂰 Channel paths
򐂰 Storage subsystem cache
򐂰 Device size and rotational speed
򐂰 Logical to physical mapping
򐂰 In summary
Pacing
Standard I/O pacing is 15 tracks per I/O operation. TDMF samples device activity every 30
seconds during the copy and refresh phases. If device or channel path activity affects the
production environment, TDMF dynamically adjusts the number of tracks read/written in a
single I/O operation. You can also use reverse pacing, which is designed to be used when
moving a volume with high activity. This option starts at one track per I/O operation and, if
activity allows, increase the number of tracks per I/O dynamically.
User-specified pacing allows you to determine the number of tracks read/written in a single
I/O operation. The values allowed are five, three, and one track per I/O. This value might or
might not be static depending on whether pacing is specified.

If an operating system is heavily used, the number of available real storage frames can be
limited. TDMF tracks this value so that it does not affect production from that standpoint. If
real storage pacing is started, I/O pacing is automatically affected.
To speed up the migration, you can select Full Speed Copy. This option allows the system to
use two buffers, increasing speed by around 45% on normal systems. See the Migration
Recommendations section of the TDMF Installation Manual for more detail.
You can also adjust the number of concurrent volumes. The current release of TDMF allows
up to 512 volumes to be coded per session. However, the number of volumes multiplied by
number of LPARs must not exceed 2048. Set this with the Number of concurrent volumes
option. Another option that can be used with the number of concurrent volumes is
Active-in-Copy. This option saves time in the overall session.
The number of concurrent running volumes can be changed dynamically using TDMF monitor
item 2 as shown in Figure 9-28. Select the field and change it, then press Enter to activate the
change. If you reduce the number, the running volumes continue copying until the end of the
volume. New volumes start when the new defined value of volumes is reached.
Session Status Row 1 to 4 of 4

System: Master Softek TDMF for z/OS Version 5.3.0

ComDataSet SYS1.TDMF.ITSO.LOCAL.SYSCOM01 Sessions 01
Number of volumes migrating 004 Number of concurrent volumes 002
Number of volumes complete 000 Number of volumes waiting 002
Requested Volume Device Group --- Migration --- - Error Info - Sync
Action Serial Number Name Status Type System Message Goal
__ RS6000 6000 Copy Swap A 005

XX8100 8100
__ RS6001 6001 Copy Swap A 005

XX8101 8101
__ RS6002 6002 Inactive Swap A 005

XX8102 8102
__ RS6003 6003 Inactive Swap A 005

XX8103 8103
Figure 9-28 Changing the number of concurrent volumes dynamically

You can manage a single volume from this TDMF monitor panel. Select the volume and use
one of the listed line commands as shown in Figure 9-29.
Transparent Data Migration Facility
PF3=End PF7=Page Up PF8=Page Down
Following valid on all Systems Command

Display Migration Messages : M
Display Migration Performance Data : D
Display Volume's Offline Volume Access (OVA) Job(s) : X
Following valid only on MASTER System Command

Acknowledge Mirror Change on Volume Swap : A
Approval Approve this migration : B
Continue Volume Migration or its Group : C or CG
Confirm PPiT Recycle for Volume or its Group : R or RG
Confirm Start Volume Migration or its Group : F or FG
Confirm Synchronization for Volume or its Group : P or PG
Disallow OVA registrations for Volume or its Group : H or HG
Discontinue PPiT migrations at "Recycle" prompt : E or EG
Reinitialize Migration of terminated Volume : I
Set Goal Synchronization Goal for Volume or its Group : Z or ZG
Suspend Volume Migration or its Group : S or SG
Terminate Volume Migration or its Group : T or TG
Terminate not allowed at APPROVAL PROMPT
PF4 Display the next active session

PF6 Toggle key for displaying the session options
PF10 Immediately display the session progress monitor
PF11 Toggle key for filtering volumes
Figure 9-29 TDMF line commands
Number of sessions
There is no limit to the number of TDMF sessions that can be started. The determining factor
is the number of available resources each LPAR has that is involved in the migration.
Factors external to TDMF

When TDMF is moving data, it uses processor resources and peripheral resources such as
channel paths and storage subsystem resources. These resources also determine how fast
data is moved.
Channel paths
TDMF interacts with the Input/Output Supervisor, which manages transport mechanism or
channel paths. Various channel paths that are in use today, from older technologies such as
parallel paths to newer ones such as FICON. Each type of channel has a specific limit as to
the amount of data that can be transported.

Table 9-1 provides raw and approximate throughput values for the types of channel paths. On
most systems, FICON is not the bottleneck.
Table 9-1 IBM S/390® channel throughput for data migration

Channel Type Raw throughput Approximate throughput
Parallel (gray) old 3.0 MBps 2.3 MBps
Parallel (blue) old 4.5 MBps 3.8 MBps
ESCON 20 MBps 12 MBps
FICON 100 MBps 85 MBps
Storage subsystem cache

The size of cache in a storage subsystem has increased over the years. However, when a
migration or migration service is being performed, it frequently involves a technology refresh.
Older technology has smaller cache sizes compared to that of new technology. Although
TDMF issues the channel command Inhibit Cache Load during the read operation, some
storage subsystems do not honor this command. Therefore, while cache is being used by the
production environment, there is the additional load of the data being moved by TDMF.
Device size and rotational speed

Over the years the size of devices has increased along with the rotational speed of the
devices. However, you often migrate off older technology, which is generally slower. Know the
source subsystem and all the intricacies that go with it.
Logical to physical mapping

In an RAID-1 environment, there are multiple logical devices that are mapped to a single
physical device. An RAID-5 environment means that there are multiple logical devices
mapped across multiple physical devices. Although these new devices are much faster than
older technology, there is still only one head-disk assembly per physical device. Moving too
many logical devices at once can be just as devastating to a production environment as
moving data without specifying pacing. However, if Solid State Disks (SSD) are installed, the
internal interfaces can be the bottleneck instead of the disks themselves.
In summary
The best migration practice is to plan. Every migration initiative, especially large migrations,
should be 90% planning and 10% implementation. Understanding the production
environment and when the heaviest loads occur is a part of panning, along with the impact of
old and new technology. Remember, data can be moved only as fast as the slowest device or
slowest channel allows.
9.3.13 Terminating a TDMF session

With the TDMF TSO Monitor or Batch Monitor, you can terminate a specific volume pairing,
volume groups, or all volumes in a session dynamically. A termination request for a volume
pairing might take up to a minute to be processed by TDMF. If the master and all the agent
systems are active in the same IBM Parallel Sysplex®, the interval is less.

Under extreme conditions, you can use the MVS CANCEL command. If the master system
fails leaving an agent system active, allow a 15-minute interval to allow the agent system to
shut down automatically. If it does not, cancel the agent system job manually.
If a TDMF session still hangs after MVS Cancel, use Display GRS,C to determine whether
canceling another job might break the deadlock. When a TDMF job is canceled, the program
signals the other systems in the session to terminate the active volumes and clean up
resources. Canceling for a second time, or using the MVS Force command, bypasses this
recovery and causes the TDMF RTM Resource Manager routine to receive control. Canceling
this way might result in migration volumes being varied offline and boxed as part of cleanup. If
you issue a Force or second Cancel command, run the TDMFCLUP program as described in
the chapter “Batch Utilities” of the TDMF Installation Manual.
Important: During an error, avoid issuing a CANCEL command using SDSF against
TDMF master or agents. Instead, use the TDMF monitor option 2 (interaction with TDMF)
to terminate a migration session.

10
Chapter 10. z/OS Dataset Mobility Facility

z/OS Dataset Mobility Facility (zDMF) is a host-based (mainframe) data migration facility that
migrates data at the logical data set (extent) level between sets of volumes. All data
movement is accomplished without application downtime.
This chapter outlines the steps needed to conduct a successful data migration project using
zDMF. It also describes the standards and provides guidelines when using zDMF for a data
migration project.
zDMF is a data set level migration tool that updates the information in the ICF catalog and
interacts with SMS. Therefore, there are sections that address its role in the migration
process, detailed explanations of the environmental requirements, and reference materials.

򐂰 z/OS Dataset Mobility Facility overview
򐂰 zDMF migration process
򐂰 Important zDMF-related topics
򐂰 Example migration in a shared DB2 test environment

10.1 z/OS Dataset Mobility Facility overview
The task of combining volumes into new, large volume capacity storage subsystems is
difficult and time-consuming. Historically combining volumes required taking
revenue-generating and often mission-critical applications offline for the duration of the
migration. z/OS Dataset Mobility Facility (zDMF) allows you to migrate and consolidate data
sets while minimizing or eliminating application downtime.
zDMF moves data sets without disrupting access to applications during the migration. zDMF
also provides the following advantages:
򐂰 Makes it easy to combine smaller capacity volumes to better use storage
򐂰 Helps to increase available system resources
򐂰 Supports implementation of tiered storage
򐂰 Improves application performance for service level compliance
򐂰 Maintains application availability during data set level migrations
򐂰 Allows continued growth of important or new applications
򐂰 Reduces storage platform total cost of ownership (TCO)
In addition, allocated data sets can be migrated with a dynamic swap of metadata. Table 10-1
shows the capabilities of zDMF.
Table 10-1 zDMF capabilities

Key factors Description zDMF capability
Dynamic swap Automatically redirects I/O to a Allows data migration without

new location after diversion disruption.
completes.
Flexible migration options Provides control over the I/O rate Allows online migration activity
for data migration reads/writes. while maintaining optimal
application performance and
service levels.
Data set grouping Allows a group of data sets to be Provides easy management of
collectively migrated. large migrations.
Robust interface Fully functional ISPF windows. Allows easy configuration,

monitoring, and operations.
Resilient architecture Current state of migration is Migration process is

maintained at all times, regardless “checkpoint” restartable in the
of controlled or non-controlled event of server shutdown and
shutdown, to ensure restart.
continued/recoverable migration.
10.2 zDMF migration process

This section outlines the steps needed to conduct a successful data migration project using
zDMF. zDMF is designed to move of allocated data sets that require non-disruptive data
migration.

10.2.1 zDMF migration steps
After the installation, use the TSO Monitor to establish a process that performs all steps of a
migration process. The sequence of steps involved in migration using zDMF is described in
Table 10-2.
Table 10-2 zDMF migration steps

Migration steps Description
Group definition The data sets to be migrated are defined in a

migration group using the TSO-based zDMF
monitor.
Activating a migration group Activating a migration group initiates the data

migration process for that group.
Copy phase Data is asynchronously copied from the source

data sets to the target data sets defined in the
migration group.
Synchronization phase All final differences between source and target

data sets in a migration group are synchronized
and the group is prepared for mirroring.
Mirror phase The migration group is put into a state of

synchronous mirroring.
Diversion phase The actual logical relocation of data sets occurs.

Source and target data sets, along with their
metadata, are modified and all I/O activity is
redirected to the new location.
Completion phase (post migration) Although the metadata has been modified,
applications that were active before diversion
continue to have their I/O redirected until they
de-allocate the data set. An application bounce
might be required in a scheduled window.
Post-Completion phase After the migration is complete, the source data

sets and the storage resources they are on will
need to be cleaned up.
10.2.2 Planning the migration
Important: The planning phase is the most vital phase for a successful migration.
This section outlines the generic steps, personnel functions, and tasks that pertain to the
planning phase. The high-level process consists of the following steps:
1. Establish a migration management team that consists of:
– Primary Migration Manager
– Alternate Migration Manager
– Account Coordinator
– Security Coordinator (if required for sensitive data)
– Technical Lead Coordinator
2. Announce the migration with a notification that aligns with the application requirements.
3. Size the target configuration to match the space requirements of the source configuration.
Chapter 10. z/OS Dataset Mobility Facility 435

4. Create a migration team to perform the pre-migration tasks listed in the following section.
The migration team should include a technical data migration team trained to use zDMF
and any other necessary utilities (that is, HSM, DFDSS, and SMS Redirection). This
technical team can also act as migration mentors.
10.2.3 Pre migration tasks

Perform the following tasks before starting a migration that includes zDMF:
1. Make sure that the account coordinator identifies the owners of the data. Inform the
owners about the migration and update them when required.
2. Inform the security and compliance groups about the migration if you are migrating
sensitive data.
3. Use the Pre-Migration Planning Checklist to ensure that all of the pre-migration planning
steps are executed.
4. Run the IDCAMS EXAMINE command with the INDEXTEST and DATATEST parameters.
Verify that the source catalog has the appropriate free space and structural integrity to
accommodate growth outside the scope of the migration.
5. Create a dedicated zDMF user catalog (see Example 10-1). Connect this catalog to the
same systems as the source catalog. Place the catalog on a non-SMS management
volume that does not contain customer data.
Example 10-1 zDMF user catalog

Source Catalog = ICF.APPL.UCAT / Connected to System A and B
Target Catalog = ICF.ZDMF.UCAT / Connected to System A and B
6. Create and relate the following aliases to the proceeding target catalog:
Define HLQs (high-level qualifiers) that are not currently used such as X001, X002, and
X003.
7. Create or add the new target volume SMS Storage Group Configuration. The new storage
capacity must be the same or greater than the source storage. Consideration must also be
given to any multivolume data sets.
8. Add these volumes to the appropriate storage group in ENABLE status as soon they are
available.
9. Create a list of the source volumes that contain data to be moved and the associated
target volumes or storage group.
10.Change the status of the source volumes to DISNEW to accommodate SMS redirection.
11.Establish a naming standard for the new target data sets (work data sets) based on the
alias created in step 6.
12.Identify the time frames that you want to move between the phases. Consider DSN and
Group activity, and fallback requirements when scheduling. In particular, schedule around
heavy I/O loads like batch work or db2 reorganizations during the migration phase.
Remember: Some of these steps might not be applicable because of your internal policies
on data migration.

10.2.4 Migration phase using DFSMS, HSM and utilities, or both
This section introduces the use of utilities to migrate data sets that are not in a persistent
allocation status. You use these utilities to identify the zDMF candidates.
DFSMS overview
Data Facility Storage Management Subsystem (DFSMS), or SMS in short, is a software suite
that automatically manages data from creation to expiration. DFSMS provides the following
functions:
򐂰 Allocation control for availability and performance
򐂰 Backup/restore
򐂰 Disaster recovery services
򐂰 Space management
򐂰 Tape management
DFSMS consists of DFSMSdf (an element of z/OS) and DFSMSdss, DFSMShsm, and
DFSMSrmm (features of z/OS).
Because DFSMS manages data creation, you can control the volume selection of these data
sets. Use ISMF or a VARY command to establish criteria to eliminate allocation of new data
sets on any volume in an SMS subsystem. This status is known as DISNEW, and disallows all
new allocations.
When starting a migration that includes moving all the data sets that populate a volume, the
source volumes must be placed in a DISNEW status. However, verify using ISMF that the
storage pool containing the volumes has enough free space so the restriction can be enabled.
Table 10-3 shows the components that address data set allocation.
Table 10-3 Data set allocation

Data set components Description
Data class Addresses the attributes of a data set.
Storage class The performance level required for the data set that is in direct
alignment with a specific service level agreement (SLA).
Management class Addresses the appropriate backup and migration cycle of the data
set.
Storage group A volume or group of volumes designated for specific data sets
based on predetermined criteria such as size, use, or type.
SMS data redirection

A storage subsystem that is under SMS control can direct and disable allocation to specific
storage pools. In an environment that has a high delete and reallocation activity, move rate
data by disabling one volume and enabling a different one in the same subsystem. This
process is known as “SMS Data Redirection.”
Because SMS Data Redirection is done on a data set level, zDMF is a good solution for data
sets that are not deleted on a regular basis. The first phase when migrating all the data sets in
an SMS environment is to have SMS move that data using SMS Data Redirection. zDMF
enables a scheduled data set movement process, which takes less time. It also allows you to
use performance enhancements from the new device for those data sets not moved using
SMS.

Note: Using SMS Data Redirection greatly decreases the deletion and recreation of data
sets during the zDMF migration process.
The DFSMShsm phase

Data sets that are not deleted and reallocated on a regular basis, but are not continuously
allocated by an online application need special attention. Despite these data sets being
frequently in use, they include a window of opportunity that can accommodate a migration.
During this time, use either Hierarchical Storage Manager (DFSMShsm) or a standard utility
to perform the migration.
When a new device type is replacing an old one, data sets must be moved from the old device
to the new one. HSM allows you to perform such conversions of a volume with a single
command. The command to convert a level 0 volume migrates all the data sets not in use on
the source volume to migration level 1, then recalls them. This command allows you to clear a
DASD migration volume so it can be replaced with a different type of storage device.
For example, to remove volume SG2002 from the system and replace it with a new volume,
you must perform the following steps:
1. Prevent new allocations to volume SG2002.
To prevent new allocations to the volume, change the status of the volume in the storage
group. For example, you can establish the volume status in the storage group definition:
STORAGE GROUP NAME: STGGP2
VOLUME SERIAL NUMBER: SG2002
SMS VOL NAME STATUS
SG2002 ===> DISNEW
2. Move the non-allocated data sets so DFSMShsm can process from the volume using the
following HSM command:
MIGRATE VOLUME(SG2002 MIGRATE(0)) CONVERT
DFSMShsm migrates all of the non-allocated (NOT IN USE) data sets that it can process
off volume SG2002 and recalls them to other SMS-managed volumes. The DFSMSdfp
automatic class selection (ACS) routine selects the volume for data set recall.
3. Remove the allocated data sets that DFSMShsm cannot process from the volume. To
move any allocated data sets that DFSMShsm cannot process, use zDMF.
Note: Data Set Services (DFSMSdss), can also be used in this phase instead of HSM.
However, HSM does not require the creation of job control language (JCL).
Standard utilities
A successful migration requires a combination of various utilities that complement each other
and enhance the migration process. If there is a window of opportunity to accommodate this
migration, any data set migration utility, such as DFSMSdss, can be scheduled to be run.
10.2.5 zDMF migration phase

After the source configuration is placed in DISNEW, the only data sets remaining on the
candidate volumes are true zDMF candidate data sets. zDMF is useful for any migration
project that would normally include the use of conventional utilities requiring a planned
application outage.

zDMF facilitates a scheduled data set movement process. This process provides these
benefits:
򐂰 Takes less time
򐂰 Allows use of performance enhancements from the new device
򐂰 Provides a utilization rate that provides a better return on investment
Tip: zDMF can be used to move any data sets that are not included in the documented
restriction list. The phased approach option can be used if needed due to user
requirements.
zDMF data set selection planning

zDMF includes many data set selection criteria, from a single data set to thousands of data
sets and multiple groups. When using selection criteria that go beyond a fully qualified data
set name, use a listing utility that uses the same selection criteria as zDMF.
The data set selection classifications can be used as selection criteria for zDMF candidates
and to identify data sets to be excluded (for example, WILDCARD selection).
The selection criteria are a key factor for data set level migration and must be included in the
migration planning process. Data set selection criteria can differ for each subsystem, which
might require use of reporting utilities.
The following are important warnings related to this process:

򐂰 To prevent any new allocation in an SMS configuration, the SOURCE volume must be in
the DISNEW status.
򐂰 Data sets that will be deleted and recreated in the zDMF migration cycle should be
excluded and handled by DFSMS Automatic Class Selection (ACS) Routines.
The following are also important factors in zDMF data set selection planning:
򐂰 Review the GDG data sets for the delete activity cycle. If the cycle is shorter than the
zDMF migration cycle, allow them move themselves.
򐂰 Change non-SMS-managed volumes that are mounted with a STORAGE status to
PRIVATE.
򐂰 The storage subsystem must have sufficient space to accommodate normal processing.
This might not be the case for data set relocation or workload balancing.
򐂰 Identify data sets with primary space 0 (SPACE=(CYL,0,100)) because they will not be
migrated.
򐂰 The security profile associated with a particular data set might prevent you from moving
the data set.
򐂰 When moving a data set for performance reasons, be sure that the new target volume
does not already have any performance constraints.
򐂰 When doing volume consolidation, be sure that you are not selecting multiple source
volumes that have performance constraints.
򐂰 Try to schedule the completion phase to take place as part of a pre-planned outage.
򐂰 Identify data sets that have a specific volser dependency. Moving data sets that have a
volser dependency without including the new volser can cause a problem.
򐂰 Identify all the volumes associated with a multi-volume data set. The same number of
target volumes are required.
򐂰 Use the Migration Planning Checklist to establish tasks, assignments, and status.

Important: When using TDMF or zDMF, make sure that catalog volumes are moved in a
separate window. In addition, make sure that catalog maintenance is not scheduled for a
basic catalog structure (BCS) associated with data sets that are being moved.
Restrictions
The following data sets are not supported by zDMF:
򐂰 Data sets cataloged in the master catalog or on any system resident volumes should not
be included in a migration.
򐂰 Data sets used by the operating system such as LINKLIST APF authorized, page, JES2,
and JES3. These data sets are not supported because they might not divert properly,
causing serious system outages.
򐂰 Virtual Storage Access Method (VSAM) with IMBED, KEYRANGE, and Replicate
parameters. For more information, see the Installation and Reference manual at:
򐂰 Catalogs.
򐂰 ISAM.
򐂰 Individual PDS members.
򐂰 HFS data sets.
򐂰 Page and swap data sets.
򐂰 Temporary (&&) data sets.
򐂰 Undefined data sets.
򐂰 Data sets allocated with no primary space or utilized space.
򐂰 Volume-specific data sets, including volume table of contents (VTOC), VSAM volume data
set (VVDS), and volume table of contents index (VTOCIX).
Both control unit types must be of an equal or higher type.

Example:
2105 -> 3990 will not be moved
3990 -> 2105/2107 will be moved
2105 -> 2105/2107 will be moved
The following data sets are flagged and do not go into Diversion phase until all associated
applications are terminated on all participating systems. For more information, see the
“Migration scheduling techniques” section in the Installation and Reference Manual.
򐂰 VSAM record level sharing data sets
򐂰 BDAM data sets that require absolute track
Data migration classifications

The following are the classifications used with data migration:
򐂰 Manage the use of source DSN masks to ensure that the population of data sets to be
migrated by any one group meet the following limitations:
– Less than 1,500 data sets
– Less than 40,000 extents total
򐂰 Appropriately sizing a migration group reduces exposures related to ECSA, CSA, and
extended private storage consumption. It also mitigates the ENQ impact of migration
processing on catalogs and volumes.

򐂰 An EXCLUDE LIST to separate data sets into various groups. Create a MASTER
EXCLUDE LIST. This list includes identifying data sets that are:
– RESTRICTED (ISAM, HFS, and so on)
– VOL SER DEPENDANT data sets
– SOFTWARE/SYSTEM data sets that are sensitive to movement
Migration group standards

Follow these guidelines when creating migration groups:
򐂰 Create group naming standards that provide easy identification. zDMF appends a ZD to a
member name, limiting your choice to six characters.
򐂰 Create groups that you can manage based on the number of data sets.
򐂰 Retain PDS members for reference.
Migration scheduling techniques

zDMF provides capabilities that are imperative for current mainframe processing
environments:
򐂰 Non-disruptive migration capabilities
򐂰 Flexibility
With zDMF, you can respond immediately to a storage-related issue. In addition, you can
manage your migration so that you maintain data set level mirroring until one of the following
phases:
򐂰 The Diversion and Completion phases in the same window
򐂰 The Completion phase at a later time after the Diversion phase
These phases are known as Extended Mirrored phase and Non-Extended Mirrored phase.
Additionally, zDMF can suspend and resume a migration using the ISPF interface. For more
information, see the zDMF Installation Manual.
Extended Mirrored phase

The zDMF migration phases from the Activation phase to the MIRRORING phase do not
include catalog alteration. While in the Mirrored Phase the two cataloged data sets, target and
source, exist and all the I/O is directed to the original source data set. Therefore, you can stay
in the Mirrored phase to accommodate the following needs:
򐂰 Migrations that include a scheduled outage that provides enough time for the Diversion
and Completion phases
򐂰 To align with internal change control policies
򐂰 For sensitive data sets that are controlled by external compliance regulations
򐂰 To accommodate business unit and or customer requirements
Extended Mirrored phase and EDSC

The EDSC option allows you to automatically divert any data sets in the group that are not
currently allocated immediately after the group enters Mirror state. The EDSC option includes
these settings:
EDSC No: All data sets remain in Mirror state until you issue the DIVERT command for the
group.
EDSC Yes: Non-allocated data sets are diverted automatically immediately after
synchronization. They are then eligible for completion processing as though the
group itself is diverted.

Non-Extended Mirrored phase
The zDMF migration management process can be set to go from the Activation phase to the
Completion (post migration) phase without extending the Mirrored phase. This process does
include catalog alteration, diverts all the I/O to the new target data set location.
Staying in the Non-Extended Mirrored phase can be used to accommodate the following
needs:
򐂰 Migrations that require immediate relocation of a data set or data sets for performance
improvements.
Tip: When moving a data set for performance reasons, be sure that the new target
volume does not already have a performance problem.
򐂰 Technology refresh projects that include devices of larger or smaller capacity.

򐂰 Physical separation of sensitive data.
򐂰 To accommodate business unit and or customer requirements.
򐂰 To accommodate storage-related issues immediately.
10.2.6 Typical migration process scenario

This section outlines the steps of a typical storage migration. In this scenario, a technology
upgrade is required because the lease of the current storage subsystem is ending. Therefore,
data needs to be migrated from old to new storage. While performing this conversion, it was
also decided to install new higher capacity drives.
The migration metrics are shown in Table 10-4.
Table 10-4 Migration metrics

Old storage New storage Capacity
800 MOD-3 to 800 MOD-3 to 2270 GB
120 MOD-3 to 40 MOD-9 340 GB
200 MOD-9 to 200 MOD-9 1702 GB
240 MOD-9 to 80 MOD-27 2043 GB
1360 volumes to 1120 volumes TOTAL=6355 GB
The new subsystem must be installed. In this scenario, the old subsystem must be removed
within 30 days of the arrival of the new subsystem to avoid costly storage overlap. It was
quickly decided that conventional disk-to-disk copy was not feasible because of the long
application outage window required. Also, the hardware migration utility of the vendor was not
usable because of compatibility issues going from ESCON to FICON. Ultimately, a
combination of host-based utilities and products was chosen to perform the migration.
Because the environment consisted of 80-90% SMS, the operations staff felt they could take
advantage of SMS redirection and the following functions:
򐂰 DFDSS to copy non-allocated files
򐂰 TDMF for volume level migrations
򐂰 zDMF for allocated files

Selection logic
TDMF was chosen to move the volume data. However, a non-disruptive logical data set
mover was needed to accommodate migration requirements. TDMF was in use but not zDMF.
It was determined that both TDMF and zDMF can accommodate the migration needs. These
two products provide for vendor-neutral, non-disruptive migration and came with a strong
endorsement from the storage vendor.
Running the migration

It was decided to perform the most complex migration, migrating the smaller to larger
volumes, first. The migration involved the following steps:
򐂰 ICKDSF Minimal Init
򐂰 Migrating smaller to larger volumes with TDMF
򐂰 Migrating using SMS and HSM processes
򐂰 Migrating one-for-one volumes using TDMF
򐂰 Example on page 444
򐂰 Completing the migration
ICKDSF Minimal Init

Run an ICKDSF Minimal Init with a VOLID like XXucb#. This function was used rather than
initializing the new volumes with VOLIDs that corresponded to their naming convention. In
addition, they were not included in the correct SMS storage pool. No SMS updates are done
to include these volumes in the SMS pools.
Migrating smaller to larger volumes with TDMF

TDMF was used to migrate the smaller volume to a larger volume, that is, one MOD-9 to an
MOD-27. This completed one-third of the migration that required resizing without any
disruption to production. It took less than one hour to migrate 40 MOD-3 to MOD-9 volumes,
and about four hours to migrate 80 MOD-9 to MOD-27 volumes.
Because the source VOLID was being copied to the target volume, no SMS changes were
needed. Changes were unnecessary because the volume was already part of the wanted
SMS rules. Also, the VOLID naming standard is propagated to the new volume.
The TDMF migration was set up to dynamically start ICKDSF to reformat and expand the
VTOC of a volume. This function is performed when the source VTOC characteristics do not
match that of the target device.
Execute this step if you have correct settings for your VTOC size when volume was moved to
new target volume. You can also do it if TDMF (ICKDSF) can extend all volumes to a correct
VTOC size.
Use the ICKDSF EXTVTOC option only if migrating from one size device to another, or if a
volume was migrated and no REFVTOC was performed. Only indexed VTOCs are extended.
Non-indexed VTOCs, including volumes with damaged indexes, are allowed.
You can specify a specific number of tracks you want the VTOC to be or you can allow TDMF
use its own algorithm as follows. The minimum size of the new VTOC is the greater of the
current VTOC size on the source and target. The VTOC can be extended further depending
on the number of data sets on the volume:
򐂰 If the volume is less than half full, the VTOC is extended to contain the current number of
data sets multiplied by the ratio of target to source device size, plus 25%.
򐂰 If the volume is more than half full, the VTOC is expanded to handle the situation where
the target volume is full of data sets with the same average size.

If the index needs to be extended, but the VTOC does not, TDMF attempts to extend the
VTOC by one track. If the VTOC cannot be extended because there is a data set next to it,
REFVTOC runs unless there is insufficient space in the index for the additional VPSMS.
Important: Resizing works only if free space is available directly behind the VTOC. If there
is no free space VTOC cannot be resized, but the new number of cylinders are updated in
VTOC.
Migrating using SMS and HSM processes

The remaining two-thirds of the source volumes were consolidated to larger capacity
volumes. These volumes were placed in the SMS DISNEW status. This status automatically
migrates files that were deleted and reallocated during daily and weekly production runs to
the new devices. It also prevents SMS from allocating any new files or extents on the old
volumes.
Operations staff can go on to their other migration tasks and allow the normal SMS routines to
move the files. However, they did need to revisit this task to determine which files SMS did not
migrate.
HSM can also be used to migrate those data sets that are not in use using the HSM
command MIGRATE VOLUME(SG2002 MIGRATE(0)) CONVERT.
Tip: Using HSM would eliminate the creation of JCL to accommodate the migration.
Migrating one-for-one volumes using TDMF

Operations staff used TDMF to migrate the remaining disks that were one-for-one. These
disks included 800 MOD-3 to MOD-3 and 200 MOD-9 to MOD-9 DASD. These volumes also
include all volumes with data sets pointed out in the restriction list of zDMF.
Migrating using zDMF

The only migration that remains to be performed is for the files which are not reallocated by
SMS to higher capacity disks in Steps 1-4. This migration was completed by zDMF.
The files that remained were files that had not been through a processing cycle. These
included the following kinds of files:
򐂰 Weekly job runs
򐂰 Files that had been created a long time ago and never got deleted
򐂰 Files that were constantly in use such as the DB2 and CICS files
Consideration was given to using DFDSS to migrate the remaining files not in allocation. It
was determined that using zDMF, operations staff can accomplish the same thing and also
handle data sets that were in use at the time. In addition, zDMF provided the flexibility of
allowing data sets to go through allocation during this process. DFDSS, however, would have
periodic allocation issues as production continued to run.
Completing the migration

All files except the ones with persistent allocations were migrated to completion. The files that
remained in zDMF Diversion mode were identified using the TSO Monitor. Those applications
were then scheduled to be bounced over the weekend.
After zDMF completed all data set migrations, the Storage Migration project to the new
technology was complete.

An additional day was spent to verify that all data had migrated as planned and post-migration
tasks had been completed.
Migration time line estimate

The following is an estimated time line (see Figure 10-1) for the migration described
previously:
򐂰 Pre-migration: Install new storage the first week.
򐂰 Migration Step 1: ICKDSF INIT new storage.
򐂰 Migration Step 2: TDMF small volumes to large volumes.
򐂰 Migration Step 3: SMS DISNEW redirection.
򐂰 Migration Step 4: TDMF equal size volumes.
򐂰 Migration Step 5: zDMF small to large volume “files” (files not moved by SMS redirection).
򐂰 Migration Step 6: Scheduled application bounce (if required).
򐂰 Migration Step 7: Clean-up.
򐂰 Post-migration: Two weeks to remove old storage (installation, migration, and
de-installation complete in one month).
Figure 10-1 Sample migration time line
10.2.7 Migration performance and scalability

This section addresses the zDMF performance parameters that establish the user needs with
specific migration requirements and system capabilities. Guidelines for sizing an zDMF
migration group are also included.
zDMF performance considerations

zDMF uses both asynchronous and synchronous techniques to copy data from source
locations to target locations, minimizing performance impact.
For performance reasons, do not place hlq.HGZD325.GZDLLIB library in the system link list
(LNKLSTxx). If the library is placed in the system link list (LNKLSTxx), remove the STEPLIB
DD statement from the JCL before running zDMF. If you place hlq.HGZD325.GZDLLIB in
PROGxx or IEAAPFxx of SYS1.PARMLIB, use the STEPLIB DD statement.
The zDMF server must always be active and able to immediately process requests. zDMF
requires a high execution priority, and should be put in a response-oriented performance
group. The zDMF-server-started task JCL specifies the configuration PDS member that

contains the startup parameters. The zDMF server configuration PDS member is located in
hlq.HGZD325.SAMPLIB(CONFIG).
Generally, do not change default values in CONFIG member. However, you might be directed
to do so by support in some special cases.
zDMF parameters
The hlq.HGZD325.SAMPLIB(CONFIG) member has the following parameters:
򐂰 MAXIO: MAXIO determines the maximum overall number of I/O requests that can be
active at one time on the server. Because I/O buffers and control areas are allocated
based on the MAXIO value, select a number appropriate to the resources available.
Approximately 1 MB of memory is allocated for each integer you add to the MAXIO
specification. For example, you select a MAXIO value of 25, approximately 25 MB is
allocated. The memory allocation is fixed during active I/O.
MAXIO=number
Where number is the maximum number of Overall Requests that can be active at one time
on the server. The minimum value is 0. If you select 0 or do not specify a MAXIO value, the
number of Overall Requests defaults to 25.
– Default: MAXIO=25
򐂰 MAX_CHANNEL_IO: Specifies the maximum number of concurrent I/O requests that can
be issued to a channel during the Copy phase or Mirror Synchronization phase. The
MAX_CHANNEL_IO limit applies to any active I/O against a channel group, whether read
or write. If the source and target devices are on the same channel group, the
MAX-CHANNEL_IO limits the total concurrent I/O requests on the channel.
MAX_CHANNEL_IO=Requests
Where Requests is the maximum number of concurrent I/O requests. The smallest value
is 0. There is no theoretical maximum value. However, the largest practical value is the
current MAXIO value.
– Default: MAX_CHANNEL_IO=15
򐂰 MAX_DEVICE_IO: Specifies the maximum number of concurrent I/O requests that can be
issued to devices containing a data set migration pair. This limitation applies during the
Copy and Mirror Synchronization phases.
MAX_DEVICE_IO=Requests
Where Requests is the maximum number of concurrent I/O requests. A value of 1 to 5 can
be specified.
– Default: MAX_DEVICE_IO=3
򐂰 MAXTRK: This optional parameter specifies the size of I/O operations in tracks that zDMF
I/O copy operations use to transfer less than a full cylinder (one extent) of data. The
MAXTRK value is used to reduce the application response time impact of zDMF Copy
operations immediately following activation. For example, MAXTRK=5 causes zDMF to
move one extent in three I/O operations. Splitting the extent reduces the time the device is
unavailable to application I/O operations into three short windows.
MAXTRK=n
Where n is a value from 1 to 15.

– Example: MAXTRK=5
– Default: MAXTRK is not used in CONFIG (parameter marked as a comment)

򐂰 UNIDENTIFIED_SYSTEMS_ACTION: Tells zDMF to check that a zDMF Server is active
on all z/OS images that can access a shared migration volume on a standard 3990-6
compatible control unit. The possible values are:
– IGNORE causes no action
– WARNing results in warning messages (GZD4192W or GZD4194W)
– TERMinate causes the group activation to fail if all z/OS images with access to the
source volume do not have an active zDMF Server. The failure message is GZD4191E
or GZD4193E. If a target volume fails the check, it is removed from candidacy,
message GZD4195I is issued, and the activation continues as long as other target
volumes are available.
Tip: Initially run in TERMINATE mode to verify correct operations and build an
IGNORE_SYSTEMS list if needed.
This parameter works with 3990-6 control units or later (2105/2107) only. TDMF cannot check
attached systems with 3990-3 (old HW.
Important: You are responsible for running the correct number of Started Tasks.
Sizing an zDMF migration group

Serialize state changes when multiple groups are used to migrate a population of data sets.
To avoid possible application performance impact and deadlock exposure, ensure groups
progress through the Activation and Pending-Divert states in serial fashion. For activations,
ensure that each group is showing a small percent of Copy/Synchronization completed before
activating successive groups. Likewise, ensure that each group is in a state of Mirrored before
entering the divert command to put successive groups in diversion.
Application performance impact can be reduced by use of the MAXTRK zDMF server
parameter.
The MAXTRK parameter causes zDMF I/O operations to be performed using less than one
cylinder. For more information, see the “Configuring the zDMF Server Parameters” in the
zDMF Installation and Reference Guide.
zDMF storage requirements

zDMF checks the available free space in ECSA while the group is activated. If there is not
enough free space available, the group are not activated. The basic ECSA, CSA, and
Extended Private storage requirements for zDMF are as follows:
򐂰 A global area is acquired that needs X'698' (1,688) bytes of fixed ECSA.
򐂰 Operating system interface modules are loaded into fixed ECSA, and require at least
X'DAA08' bytes (874.5 KB).
򐂰 The PC routine is loaded into pageable ECSA, which currently requires X'B3F00' bytes
(720 KB).
򐂰 The storage pools (all fixed ECSA) are initialized with X'480000' bytes (4.5 MB). When
groups are mirroring or diverting, the storage pools expand depending on the complexity
of the channel programs.
򐂰 When a group is promoted and activated, the parsing routine uses ECSA to store images.
򐂰 For each active group, the following are true (in fixed ECSA):
Control blocks for which space is not freed:

– One DWGRP block, X'A8' (168) bytes, until the next IPL, or the product is shut down.
– One DWGDEV block for each source and target device, X'F8' (248) bytes.
– A sparse array for extent look-up from each device block. Minimum size is X'400' bytes
(1 KB).
Control blocks for which space is freed when the group completes:
– One QCDSN block for every data set and each individual VSAM component or AIX,
X'2B8' (696) bytes.
– One DWEXT block for each data set extent, X'68' (104) bytes.
Note: At least 0.85 MB is resident in fixed ECSA until the next IPL for all components,
except for the PC routine and possibly the storage pools.
Security
Protect zDMF using IBM RACF®, ACF2, or other security utilities to prevent use by
unauthorized personnel. Use of zDMF by unauthorized personnel might result in the
inappropriate transfer of a data set out of a specified isolated environment.
Volume checklist
Be sure that the TARGET configuration is added to the appropriate storage group and all the
volumes remain in DISALL status until you are ready to start the migration. The volume status
then changes to ENABLE.
Be sure that all of the TARGET volumes meet the following requirements:
򐂰 Are initialized for SMS (ICKDSF STORAGEGROUP)
򐂰 Contain a configuration large enough for a VVDS
򐂰 Contain an INDEX VTOC
Because the INDEX VTOC is required on an SMS-managed volume, ISMF can be used to
identify a disabled or broken INDEX VTOC.
Be sure that all the SOURCE volumes are in DISNEW status. This includes volumes with a
similar capacity configuration even if they do not contain data sets that are scheduled to be
moved. An example is all mod 3 in the same STG GRP. In an SMS configuration, this status
prevents any movement back to a DEVICE TYPE that is scheduled for removal. If DISNEW
status was enabled using an SMS VARY command, the next ACS TRANSLATE and
VALIDATE can reverse that status.
Tip: SNAP requires an INDEX VTOC on the volume. Also, the VTOCIX must match the
volser or the migration fails.
In this example, the migration continues:

SYS1.VTOCIX.TD538D at volume TD538D
In this example, the migration fails:

SYS1.VTOCIX.SPMS19 at volume YPMS19
Migrating software vendor data sets

The zDMF data set selection criteria and other restrictions are based on specific access
methods and data set types. These restrictions are clearly outlined in the zDMF product
documentation.

However, you must first consider additional software vendor restrictions on some product data
sets. These restrictions are essential to preserve data integrity and product-related
functionality. Additionally, before running the zDMF migration, make sure that you have
documentation of user exits, product customization, and cross-application dependencies
unique to your environment.
zDMF can move any data sets that conform to supported data set lists and other restrictions.
However, consider additional application restrictions and consult with the person responsible
for the application. Also, keep in mind that each installation, although apparently similar, can
have subtle differences that might jeopardize the migration.
10.2.8 Post-migration
The creation of a separate user catalog for the zDMF target data sets expedites the
post-migration cleanup process. Additionally, the renamed source data set provides content
verification that you can use for a post-migration audit trail. The following are the steps for
various post-migration scenarios:
For a full volume migration:

򐂰 Delete the aliases related to the zDMF user catalog.
򐂰 Delete the zDMF user catalog.
򐂰 Initialize the old source volumes.
For a selective data set migration:

򐂰 Delete the work data sets using IDCAMS:
DELETE (X00x.*)
򐂰 Delete the data sets using ISPF 3.4 or ISMF.
Or use the zDMF monitor function Z Generate Source DSN Cleanup JCL in front of a
completed group delete all work data sets.
򐂰 Delete the aliases related to the zDMF user catalog.
򐂰 Delete the zDMF user catalog.
For a full volume migration with post migration verification:

򐂰 List the original zDMF candidate data sets that can be found on the original volumes.
If the volume contained data sets that are either restricted or invalid, they must be factored
in the original data set count.
򐂰 Based on the zDMF selection criteria, list the new data sets using IDCAMS.
򐂰 Compare the data sets. Because both the target and source data set configurations
consist of cataloged data sets, the contents of either can also be verified.
For a data set migration with post-migration verification, perform the following steps:
򐂰 List the original zDMF candidate data sets using any listing utility and the same zDMF
selection criteria.
If the volume contained data sets that are either restricted or invalid, they must be factored
into the original data set count.
򐂰 Based on the zDMF selection criteria, list the new data sets using IDCAMS.
򐂰 Compare the data sets. Because both the target and source data set configurations
consist of cataloged data sets, the contents of either can also be verified.

Tip: After the migration group completes and the verification process is satisfied, delete
the migration group from the zDMF database. Use the zDMF TSO Monitor, Option 2 -
Interact With Promoted Groups, to delete the group.
When a migration group is deleted, the zDMF server performs the following tasks:
򐂰 Ensures that the migration group is in a state that allows it to be deleted.
򐂰 Removes the migration group from internal z/OS storage (memory). Removing the
group frees valuable system memory.
򐂰 Deletes the migration group from the zDMF database.
As part of the post-migration activities, address the following concerns:

򐂰 Create a clean-up schedule to determine how long you want to keep the zDMF target data
sets.
򐂰 Determine how you want to treat the zDMF data sets:
– Back up before delete
– Delete without backup
– Delete the BCS and Init the volumes
– Content verification for post-migration audit
10.3 Important zDMF-related topics

This section includes other important concepts that you need to understand in the context of
an zDMF migration.
10.3.1 Understanding the catalog structure

Because zDMF is a data set-level migration tool that includes updating the information in the
ICF catalog, you must have a basic understanding of the ICF catalog structure. This section
provides a high-level outline of the structure to help engineers do the following tasks:
򐂰 Perform diagnostic tests for the pre-migration process
򐂰 Identify the cause of migration problems resulting from erroneous entries
The ICF catalog records are stored in two components:

򐂰 The basic catalog structure (BCS): BCS is considered the catalog. The BCS is a VSAM
key-sequenced data set (KSDS). Its primary function is to point to the volumes on which a
data set is located. The BCS is created when either a user or master catalog is defined
using access method services (AMS). A BCS does not have to be on the same volume as
the data set it references. There can be more than one BCS on a volume.
򐂰 The VSAM volume data set (VVDS): VVDS is considered an extension of the VTOC. The
VVDS is a VSAM entry-sequenced data set (ESDS). It contains the information required
to process VSAM data sets and, in a Storage Management Subsystem (SMS)
environment, it contains the volume-related information for non-VSAM SMS-managed
data sets. There is one VVDS on each DASD volume that contains a VSAM or
SMS-managed data set cataloged in an ICF catalog. The VVDS is always on the same
volume as the data set it references.
The relationship between the BCS and the VVDS is many-to-many. A BCS can point to
multiple VVDSs, and a VVDS can point to multiple BCSs.

Any volume containing a BCS also contains a VVDS because the BCS is itself a VSAM data
set.
The records in both the VVDS and BCS are made up of variable-length cells and subcells.
The two cell types that are often referred to are the VSAM volume record (VVR) and the
non-VSAM volume record (NVR), which are both held in the VVDS. The VVR contains
information relating to VSAM data sets. The NVR contains information relating to non-VSAM
SMS-managed data sets.
Most data sets have entries in only one VVDS. However, multivolume data sets have entries
in the VVDS of each volume they are allocated to (type Q for a secondary record).
Table 10-5 shows the structure of the ICF catalog.
Table 10-5 ICF catalog structure

Information VSAM data set SMS-managed Non-SMS-mana Uncataloged
non-VSAM data ged non-VSAM non-VSAM data
set data set set
Volume BCS BCS BCS n/a
Data set type BCS BCS BCS n/a
Association BCS BCS BCS n/a
Ownership BCS BCS BCS n/a
SMS class info BCS & VVDS BCS & VVDS n/a n/a
Data set VVDS VVDS & VTOC VTOC VTOC

attributes
Extent VVDS & VTOC VTOC VTOC VTOC

description
Catalog name VVDS VVDS n/a n/a
10.3.2 BCS record types

The first cell in each record has a cell type field, which is also the record type (or ID). The
record type can be identified using the DIAGNOSE command. The following are the possible
record types and their one-character identifiers.
BCS ID record types:

򐂰 A non-VSAM data set
򐂰 B generation data group
򐂰 C cluster
򐂰 D data component
򐂰 E VSAM extension record
򐂰 G Alternate index
򐂰 H GDS
򐂰 I index component of a cluster
򐂰 J GDG extension cell
򐂰 L library
򐂰 R path
򐂰 T true name
򐂰 U user catalog connector

򐂰 W volume
򐂰 X alias
VSAM Volume Record (VVR):

򐂰 Z primary record
򐂰 Q secondary record and data sets allocated with IMBED
򐂰 N Non-VSAM Record (NVR) for non-VSAM data sets
10.3.3 Catalog diagnostic recommendations

The BCS, VVDS, and VTOC each supply a portion of the information required to process a
data set. Damage to any piece or a mismatch between them can prevent you from accessing
your data even if there is no problem with the data set itself. Most data sets have entries in
only one VVDS or VTOC. The exception is a multivolume data set, which has entries in the
VVDS or VTOC of each volume it occupies.
Because the zDMF data movement process includes catalog alteration, you must start with a
functional catalog. In addition, verify that the candidate data sets are correctly cataloged on
the system or systems to which zDMF requires access. Use SSMzOS or IDCAMS to perform
the required diagnostic tests.
BCS/VVDS/VTOC synchronization
A SSMzOS catalog diagnostic checks the following items:
򐂰 Data structure
򐂰 Index integrity
򐂰 BCS/VVDS/VTOC synchronization
The catalog diagnostic tests can also generate IDCAMS control cards to fix the errors
detected. Specifically, the IDCAMS control cards for DEFINE RECATALOG can be generated
to recreate missing VSAM and non-VSAM catalog entries. The catalog commands and their
actions are shown in Table 10-6.
Additionally, there are diagnostic routines to detect and remove superfluous catalog entries.
These diagnostic routines can be directed to your entire catalog environment, or assigned to
a specific catalog, volume, or object (catalog, VVDS, or volume). Specific objects are selected
using keywords that allow specific and generic parameters. When SSMzOS catalog-related
functions are run on a scheduled basis, these steps establish consistent catalog
synchronization. This synchronization is important when running post-catalog recovery
diagnostic tests and explaining erroneous entries in a BCS or a VVDS.
Table 10-6 Catalog command and action summary table

Command Action
DELETE NOSCRATCH Removes only BCS entries. VVDS and VTOC are
not affected.
DELETE TRUENAME Removes unrelated truename record if the

associated cluster record does not exist.
DELETE VVR or DELETE NVR Removes unrelated VVR or NVR. Also removes
VTOC entry, if present.
DEFINE RECATALOG Creates a BCS entry pointing to existing VVRs or

NVR only in the BCS they name.

Command Action
DEFINE NONVSAM Creates BCS entry pointing to

non-system-managed non-VSAM data set. This
command does not verify the existence of a data
set on the named volume.
The EXAMINE command

To guarantee that a BCS and an associated backup are structurally sound, the IDCAMS
EXAMINE command must be included in the scheduled BCS backup. This command traces
failures caused by unsynchronized entries in the catalog, and shows if the catalog might be
structurally unsound. Structural problems affect BCS characteristics as a VSAM
key-sequenced data set, not as a catalog. Test both the index and data components of a
BCS, and identify the amount of free space remaining in the BCS.
Sample JCL
This section lists sample JCLs.
Diagnostics: BCS/VVDS/VTOC synchronization

The following JCL provides VVDS to BCS synchronization:
//STEP01 EXEC PGM=IDCAMS
//SYSPRINT DD SYSOUT=A
//VOLDD DD VOL=SER=PROD01,DSN=SYS1.VVDS.VPROD01,NAME,DISP=OLD
//SYSIN DD *
DIAGNOSE VVDS INFILE(VOLDD) COMPAREDS(ICF,BCS.UCAT)
ERRORLIMIT(200)INDEXTEST DATATEST
/*
For VTOC to BCS synchronization, see the SSMzOS catalogs functions.
Diagnostics: IDCAMS DIAGNOSE

This example shows how to determine whether the index component of your catalog has
structural errors:
//EXAMEX1 JOB
//STEP1 EXEC PGM=IDCAMS
//SYSPRINT DD SYSOUT=A
//SYSIN DD *
EXAMINE NAME(ICFCAT.V338001/MASTRPW) -
ERRORLIMIT(200)INDEXTEST DATATEST
/*
The EXAMINE command is used to analyze the index component of an integrated catalog
facility catalog. Its parameters are:
򐂰 NAME: Specifies the catalog name and its master password. The catalog must be
connected to the master catalog.
򐂰 INDEXTEST: Specified by default.
򐂰 ERRORLIMIT(0): Suppresses the printing of detailed error messages.
Sample output:
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
Then following is an example of the IDCAMS control statements that can be used for
VVDS and BCS diagnostics.
Compares the forward pointers from the BCS to the VVDS:

Compares the backward pointers from the VVDS to the BCS:
Tip: INCAT is the DD statement that includes the BCS name
10.3.4 Using Global Resource Serialization (GRS)

In a Global Resource Serialization complex, programs can serialize access to data sets on
shared DASD volumes at the data set level. A program on one system can access a data set
on a shared volume while other programs on any system can access other data sets on the
same volume. Serialization can therefore reduce contention for these resources and minimize
the chance of an interlock occurring between systems. This serialization ends the need to
protect resources by job scheduling.
Because global resource serialization maintains information about global resources in system
storage, data integrity is maintained during a system reset while a reserve exists. A global
resource serialization complex also allows serialization of shared logical resources. These
resources do not have to be directly associated with a data set or DASD volumes. An ENQ
with a scope of SYSTEMS can be used to synchronize processing in multisystem
applications. Combining the systems that access shared resources into a global resource
serialization complex can solve the problems related to using the Reserve macro.
How global resource serialization works depends on the z/OS operating system level. It also
depends on whether the cross-system coupling facility (XCF) environment the systems are
running in is sysplex or non-sysplex.
The complex consists of one or more systems:

򐂰 Connected in a ring configuration through:
– Global Resource Serialization-managed channel-to-channel (CTCs) adapters
– XCF communication paths (CTCs)
– Signaling paths through a Coupling Facility
򐂰 Connected to a coupling facility lock structure in a star configuration through:
– Signaling paths
– XCF communication paths
– A combination of both signaling paths and XCF communication paths
No matter which configuration you choose, global resource serialization processing

(ISGENQ, ENQ, DEQ, and RESERVE requests) for resources is the same.
10.3.5 Ring complex

As stated earlier, a global resource serialization ring complex consists of one or more
systems connected by communication links. Global resource serialization uses the links to
pass information about requests for global resources from one system in the complex to
another.
The global resource serialization complex consists of every system that indicates, at IPL time,
that it is to be part of the complex. This complex is not affected by the physical configuration
of systems and links. The systems in the complex might not all be actively using global
resource serialization at any particular time. Those systems that are actively using global
resource serialization to serialize access to global resources make up the global resource
serialization ring.

Figure 10-2 shows a four-system global resource serialization ring complex. When all four
systems in the complex are actively using global resource serialization, the complex and the
ring are the same. The complex has a communication link between each system and every
other system. It is therefore a fully connected complex.
System 1 System 3
System 2 System4
Figure 10-2 Ring concept
A sysplex requires full connectivity between systems. Therefore, when the sysplex and the
complex are the same, the complex has full connectivity. Although a mixed complex might not
be fully connected, a fully connected complex allows the systems to build the ring in any
order. It also allows any system to withdraw from the ring without affecting the other systems.
The complex offers more options for recovery if a failure disrupts ring processing.
For example, if system System1 in Figure 10-2 were to fail and end its active participation in
serializing access to global resources, it would still be part of the complex. However, it would
not be part of the ring.

10.3.6 Star complex
The star method for processing requests for global resources operates in a sysplex like any
other z/OS component that uses the Coupling Facility (Figure 10-3). Unlike the ring method,
which might require you to set up XCF signaling paths for the SYSGRS group, the star
method operates well without any special definitions. There is no need to set up XCF
signaling paths between systems or to define CTC links directly to global resource
serialization.
System1
Coupling
Shared System 2 Facility
DASD (XCF)
System 3
Figure 10-3 Star concept
XCF requires a DASD data set, called a sysplex couple data set. This data set is shared by all
systems. An alternate data set can be used to facilitate migration from a ring to a star
complex. On the sysplex couple data set, z/OS stores information related to the sysplex
systems, and XCF groups such as global resource serialization. The following policies are
used to manage global resource serialization:
򐂰 Coupling facility resource management (CFRM) policy, which is required, defines how
MVS manages Coupling Facility resources.
򐂰 Sysplex failure management (SFM) policy, which is optional, defines how MVS is to
manage the system. It signals connectivity failures and IBM PR/SM™ reconfiguration
actions. Generally, use the SFM policy.
Tip: XCF is a component of the z/OS operating system that supports cooperation between
authorized programs running in a sysplex.
10.3.7 Reserve handling requirements

zDMF has specific requirements in the area of reserve handling. More specifically, zDMF
requires that all hardware reserves issued by z/OS address spaces be converted to globally
propagated ENQ requests. These reserves include system, application, and jobs.

Understanding Global Resource Serialization and Hardware Reserves
Global resource serialization (GRS) is a function that allows serialized use of resources.
Through enqueue and dequeue facilities, usage of shared resources can be guaranteed.
Resources can either be simultaneously shared across multiple users of the resource, or
granted exclusively to one user.
GRS is useful in a shared storage environment in which multiple z/OS images are using the
same storage resources. To control use of a resource, the z/OS operating system issues a
hardware reserve to the physical device. This reserve ensures that no other z/OS image in
the shared storage environment can access that device so long as the reserve is still held.
To improve system performance and through-put, products that convert hardware reserves to
globally propagated ENQ requests were developed. An example is the Multi-Image Manager
by Computer Associates. These products communicate with all systems in the shared
storage complex. They ensure exclusive usage by issuing a software enqueue on all systems
in the shared storage environment.
When using zDMF to migrate data, the following protections are provided while converting
hardware reserves to globally propagated ENQ requests:
򐂰 Ensuring the data integrity of data for data sets being migrated
򐂰 Avoiding deadly embraces in multisystem environments
Ensuring data integrity for data sets being migrated

To ensure the data integrity of data sets being migrated, hardware reserves must be
converted to globally propagated ENQ requests. In the following examples, data set DS1 is
being migrated from source Volume A to target Volume B. The applications are running on
two z/OS servers, MVSA and MVSB, and both require exclusive serialization of the resource.
zDMF impact on serialization before diversion

zDMF phases that run on a migration before diversion have no impact on resource
serialization. These phases include Group Definition, Activation, Synchronization, and
Mirroring. Because all applications are continuing to read and write data from the source
location, all serialization is targeted to the source location. For example, data set DS1 on
Volume A has all reserves issued to that location across all applications and systems for
serialization of the resource. Therefore, all I/O activity to the Mirrored data set or target on
Volume B is also serialized.
zDMF impact on serialization after diversion

During the zDMF diversion phase, hardware reserves are converted to ENQ requests to
ensure data integrity of data sets being migrated.
After entering the diversion phase, applications that remain active after diversion will continue
to serialize usage of the resource as though it still on the source volume. Applications starting
after diversion are allocated directly to the new target location and do not require any I/O
diversion activity by zDMF. These applications do not require alteration because all catalog
and volume level metadata are modified to reflect the new location of the data set.
When this scenario occurs, an exposure exists with resource serialization for the data set.
The following is an example of how this scenario can occur:
򐂰 MVSA reserves Volume A for data set DS1 which is in diversion, reads data, and prepares
to write it back. The data read occurs on Volume B because the migration is in the
Diversion phase.
򐂰 On MVSB, a new application starts and reserves volume B. The application reads data
from data set DS1, modifies it, and writes it back to data set DS1.

򐂰 During this time, the zDMF channel program for MVSA must wait because Volume B is
reserved to MVSB. After MVSB releases the reserve on Volume B, MVSA will overwrite
what MVSB wrote, and then release the reserve on Volume A.
In this scenario, a classic data integrity exposure occurs. The data is overwritten without any
indication that this has occurred.
If the reserve is converted to a global enqueue, the resource is protected by name and
remains secure. This protection remains as long as all applications protecting any associated
resource use the same name and do not rely on the actual device reserve for serialization. An
example is when other data sets are thought to be on the same volume or any logical
resource. If the reserve is not converted, the benefit of reserve protection is lost between pre-
and post-diversion applications.
Avoiding deadly embraces in a multi-system environment

The SYSVTOC and SYSZVVDS resource names are the QNAMES for resource serialization
of the VTOC and VVDS of each volume. These data sets contain all the control information
for data sets on that volume.
Data sets grow and shrink with usage. As they do, the VTOC and VVDS must be updated to
reflect the space usage on that volume. zDMF mirrors all activity for a data set on a source
volume to a target volume. When extent changes occur, zDMF must ensure that similar
changes occur on the target volume. For example, if DS1 (data set 1) on Volume A adds an
extend, zDMF must also add an extent on Volume B for the mirrored data set.
If hardware reserves are not converted for the VTOC and VVDS, hardware reserves must be
issued to both physical volumes. This can result in a deadly embrace.
To avoid this situation, these resources must be converted from hardware reserves to ENQ
requests. This conversion allows special handling code in zDMF to prevent multiple systems
from trying to exclusively acquire the same volume resources.
If these resources are not converted from hardware to ENQ requests, you might have to
cancel address spaces to eliminate the deadly embrace.
To convert the reserves, use products such as CA-MIM or GRS by IBM. zDMF provides a
stand-alone utility called Reserve Monitor. This utility detects hardware reserves and
determines whether they are currently being converted to globally propagated ENQ requests.
For more information, see “Reserve Monitor” in the zDMF installation manual.
10.4 Example migration in a shared DB2 test environment

In this example, active allocated data sets in a shared db2 environment are moved from
3390-3 to 3390-9. The following steps were performed:
򐂰 The storage group was prepared, source volumes were set to DISNEW, and new target
volumes were set to ENABLE.
򐂰 Before target volumes were added to the storage group, table spaces were allocated at
the source volumes and records include data were added.
򐂰 All records were checked.
򐂰 New target volumes added to the storage group and set to ENABLE, old source volumes
defined as “DISNEW”.
򐂰 During migration updates were running against the table spaces out of two systems.

򐂰 Groups for zDMF were created using BATCH.
򐂰 Migration was monitored using the zDMF monitor.
After migration, all data were checked for consistency.
10.4.1 Test Environment Overview

Before starting, check that all prerequisites are met as described previously in this chapter.
Figure 10-4 shows the System Managed Storage (SMS) volume status.
zDMF Test Scenario
db2 TSO
zDMF
Monitor
workload MCECEBC
CECE db2
MZBCVS2 zDMF (Started Tasks)

BVS2
Storage Group:
Communication
SG3390X1
Database
Migrate
Source
Dev: 6030-603F
VOLID: Target
RS6030-RS603F Dev: 850A-850F
3390-3 VOLID:
ZD850A-ZD850F
3390-3, SMS - DISNEW 3390-9
3390-9
SMS - ENABLE
IBM ESS 800

IBM DS8100
Figure 10-4 zDMF test environment overview
zDMF was already installed and active as shown in Figure 10-5. For performance reasons,
the zDMF communication data base and the user catalog for zDMF work data sets were on
NONSMS managed volumes.
Subsystem ZD Host Messages Row 1 to 2 of 2

SMFID SSID CPUID Last Server Cold Start Time
_ CVS2 ZDMF 05423A2064 06/24/2011 18:28:01.947

_ CEBC ZDMF 04423A2064 06/24/2011 18:28:00.568
Figure 10-5 Check zDMF systems

These data sets can be checked by using the zDMF Monitor, as shown in Figure 10-6.
Installation Options Row 1 to 26 of 26

Company : IBM Deutschland GMBH

Site . : IBM MAINZ
Site ID : 11660 Date Initialized . . . : January 10, 2011
Local Time on Messages . . : Yes
Operating Environment
IBM zDMF
zDMF VERSION : 3.2.5
zDMF COMMAND PREFIX (CPFX) : ZD
zDMF DATA BASE DATA SET NAME : SYS1.HGZD325.DB <-- DB name
zDMF DATA BASE VOLUME : ZDMFDB <-- DB vol
zDMF DB SYSTEM AUTHORIZATION : ALTER
zDMF SIMULATE DATA SET NAME : SYS1.HGZD325.SIMULATE.REPORTS
zDMF SIMULATE VOLUME : VSLNSM
zDMF SIM SYSTEM AUTHORIZATION : ALTER
SMF INTERVAL (# of minutes) : 5
SMF TYPE :
SMF SUBTYPES INCLUDED :
UNIDENTIFED SYSTEMS ACTION : IGNORE
IGNORED SYSTEMS : NONE DEFINED
USER ID : STXXXXX
CPU ID : 0004423A 2064
SCP NAME : SP7.1.2
SCP FMID : HBB7770
ETR ID : 00
LOCAL TIME : 07/09/2011 00:10:45.09
GMT TIME : 07/08/2011 21:59:40.67
LOCAL OFFSET : +02:11:04
LEAP SECONDS : +000
Figure 10-6 zDMF Monitor, Item 7
You can also check the database and user catalog with TSO 3.4 with the High Level Qualifier
(HLQ) of “CATALOG.ZDM*” (Figure 10-7).
DSLIST - Data Sets Matching CATALOG.ZDM* Row 1 of 2


-------------------------------------------------------------------------------
CATALOG.ZDMICFU.VZDM001 ZDM001
CATALOG.ZDMICFU.VZDM001.CATINDEX ZDM001
***************************** End of Data Set list ****************************
Figure 10-7 TSO 3.4 - zDMF User Catalog

The alias for the work data sets used (HLQ of XXX) was added to the zDMF user catalog.
Check for the volumes in storage group SG3390X1 and the SMS status as shown in
Figure 10-8. Source volumes start with RS603x in the volume ID, and target volumes start
with ZD850x.
SDSF ULOG CONSOLE STXXXXX1 LINE COMMAND ISSUED

COMMAND INPUT ===> /D SMS,SG(SG3390X1),LISTVOL SCROLL ===> CSR
RESPONSE=MCECEBC
IGD002I 00:34:31 DISPLAY SMS 172
STORGRP TYPE SYSTEM= 1 2

SG3390X1 POOL + +
VOLUME UNIT SYSTEM= 1 2 STORGRP NAME

RS603A 603A D D SG3390X1
RS603B 603B D D SG3390X1
RS603C 603C D D SG3390X1
RS603D 603D D D SG3390X1
RS603E 603E D D SG3390X1
RS603F 603F D D SG3390X1
RS6030 6030 D D SG3390X1
RS6031 6031 D D SG3390X1
RS6032 6032 D D SG3390X1
RS6033 6033 D D SG3390X1
RS6034 6034 D D SG3390X1
RS6035 6035 D D SG3390X1
RS6036 6036 D D SG3390X1
RS6037 6037 D D SG3390X1
RS6038 6038 D D SG3390X1
RS6039 6039 D D SG3390X1
ZD850A 850A + + SG3390X1
ZD850B 850B + + SG3390X1
ZD850C 850C + + SG3390X1
ZD850D 850D + + SG3390X1
ZD850E 850E + + SG3390X1
ZD850F 850F + + SG3390X1
***************************** LEGEND *****************************
. THE STORAGE GROUP OR VOLUME IS NOT DEFINED TO THE SYSTEM
+ THE STORAGE GROUP OR VOLUME IS ENABLED
- THE STORAGE GROUP OR VOLUME IS DISABLED
* THE STORAGE GROUP OR VOLUME IS QUIESCED
D THE STORAGE GROUP OR VOLUME IS DISABLED FOR NEW ALLOCATIONS ONLY
Q THE STORAGE GROUP OR VOLUME IS QUIESCED FOR NEW ALLOCATIONS ONLY
> THE VOLSER IN UCB IS DIFFERENT FROM THE VOLSER IN CONFIGURATION
SYSTEM 1 = MCECEBC SYSTEM 2 = MZBCVS2
Figure 10-8 Displaying all volumes within storage group SG3390X1 with SMS status

Statistical data for the volume allocation status showed that all new volumes were empty
(Figure 10-9).
VOLUME DEV DEV MOUNT FREE SPACE ! LARGEST FREE ! PCT FRAG
SERIAL TYPE NUM CHAR CYLS TRKS EXTS ! CYLS TRKS PCT ! FREE INDEX
****** ------ ---- ----- ----- ---- ---- ! ---- ---- ---- ! ---- -----
RS603A 33903 603A SMS X 1020 8 1 ! 1020 8 100% ! 30% .000
RS603B 33903 603B SMS X 1462 14 1 ! 1462 14 100% ! 43% .000
RS603C 33903 603C SMS X 1616 11 1 ! 1616 11 100% ! 48% .000
RS603D 33903 603D SMS X 568 7 1 ! 568 7 100% ! 17% .000
RS603E 33903 603E SMS X 1324 10 1 ! 1324 10 100% ! 39% .000
RS603F 33903 603F SMS X 996 6 1 ! 996 6 100% ! 29% .000
RS6030 33903 6030 SMS X 841 14 1 ! 841 14 100% ! 25% .000
RS6031 33903 6031 SMS X 696 6 2 ! 696 0 99% ! 20% .000
RS6032 33903 6032 SMS X 1764 10 1 ! 1764 10 100% ! 52% .000
RS6033 33903 6033 SMS X 1460 3 1 ! 1460 3 100% ! 43% .000
RS6034 33903 6034 SMS X 1002 4 1 ! 1002 4 100% ! 30% .000
RS6035 33903 6035 SMS X 697 12 1 ! 697 12 100% ! 20% .000
RS6036 33903 6036 SMS X 1762 14 1 ! 1762 14 100% ! 52% .000
RS6037 33903 6037 SMS X 1168 7 1 ! 1168 7 100% ! 35% .000
RS6038 33903 6038 SMS X 302 9 3 ! 297 0 98% ! 9% .013
RS6039 33903 6039 SMS X 704 5 1 ! 704 5 100% ! 21% .000
ZD850A 33909 850A SMS X 9974 5 3 ! 9822 0 98% ! 99% .007
ZD850B 33909 850B SMS X 9974 5 3 ! 9822 0 98% ! 99% .007
ZD850C 33909 850C SMS X 9974 5 3 ! 9822 0 98% ! 99% .007
ZD850D 33909 850D SMS X 9974 5 3 ! 9822 0 98% ! 99% .007
ZD850E 33909 850E SMS X 9974 5 3 ! 9822 0 98% ! 99% .007
ZD850F 33909 850F SMS X 9974 5 3 ! 9822 0 98% ! 99% .007
Figure 10-9 Space Information for Storage Group “SG3390X1”
10.4.2 Preparing DSN List for Test Migration

The test migration example started the migration group using BATCH JCL. A DSN list
therefore needed to be created with data set names on the source volumes. The data sets
were created using the following filter mask:
DSNCAT.DSNDBD.DBX1ST01.*

This filter mask results in 30 data sets, 15 tables, and 15 index data sets as shown in
Figure 10-10.
DSLIST - Data Sets Matching DSNCAT.DSNDBD.DBX1ST01.* Row 1 of 30

Command - Enter “/” to select action Message Volume

-------------------------------------------------------------------------------
DSNCAT.DSNDBD.DBX1ST01.IXA010A.I0001.A001 RS603A+
DSNCAT.DSNDBD.DBX1ST01.IXA010B.I0001.A001 RS6037+
DSNCAT.DSNDBD.DBX1ST01.IXA010C.I0001.A001 RS6031+
DSNCAT.DSNDBD.DBX1ST01.IXA010D.I0001.A001 RS603E+
DSNCAT.DSNDBD.DBX1ST01.IXA010E.I0001.A001 RS6035+
DSNCAT.DSNDBD.DBX1ST01.IXA010F.I0001.A001 RS6030+
DSNCAT.DSNDBD.DBX1ST01.IXA0101.I0001.A001 RS603C+
DSNCAT.DSNDBD.DBX1ST01.IXA0102.I0001.A001 RS6031+
DSNCAT.DSNDBD.DBX1ST01.IXA0107.I0001.A001 RS603F+
DSNCAT.DSNDBD.DBX1ST01.IXA0108.I0001.A001 RS603A+
DSNCAT.DSNDBD.DBX1ST01.TSA010A.I0001.A001 RS603B+
DSNCAT.DSNDBD.DBX1ST01.TSA010B.I0001.A001 RS6030+
DSNCAT.DSNDBD.DBX1ST01.TSA010C.I0001.A001 RS6032+
DSNCAT.DSNDBD.DBX1ST01.TSA010D.I0001.A001 RS6034+
DSNCAT.DSNDBD.DBX1ST01.TSA010E.I0001.A001 RS6030+
DSNCAT.DSNDBD.DBX1ST01.TSA010F.I0001.A001 RS6032+
DSNCAT.DSNDBD.DBX1ST01.TSA0101.I0001.A001 RS6038+
DSNCAT.DSNDBD.DBX1ST01.TSA0102.I0001.A001 RS603F+
DSNCAT.DSNDBD.DBX1ST01.TSA0108.I0001.A001 RS603A+
Figure 10-10 Data set name list for test migration
10.4.3 Creating BATCH JCL

Figure 10-11 on page 464 shows a JCL example for zDMF batch. Split a long DSN list into
groups of 16 data sets and start with a “SET” statement. It is helpful to use a REXX procedure
to create this BATCH JCL out of an input DSN list.
The following are the zDMF parameters used in this test example:
򐂰 DELETE_EXISTING_TARGET_DATASETS (NO): Do not delete existing work data sets
with same name.
򐂰 EARLY_DATA_SET_COMPLETION (YES): Divert and complete all non-allocated data
sets as soon as MIRROR status is reached.
򐂰 ALLOCSEQ (NONE): Move data sets as they are in list rather than sort by size.

򐂰 SPHERE (YES): Move all parts of a VSAM Cluster.
򐂰 TOLERATE_SOURCE_VALIDATE_FAILURE (YES) and
TOLERATE_TARGET_CREATE_FAILURE (YES): Do not stop the migration in case of an
allocation error. Instead, skip this data set and proceed with next.
//STRZ999 JOB (A185,SYS),'STXXXXXX IBM',

// MSGLEVEL=(1,1),NOTIFY=&SYSUID,MSGCLASS=X,CLASS=A
//* IBM user name and telefon number
/*JOBPARM S=CEBC
//BATCH EXEC PGM=GZDBAT,PARM='CPFX=ZD'
//STEPLIB DD DISP=SHR,DSN=CPAC.HGZD325.GZDLLIB
//LDMFBAT DD SYSOUT=*
//SYSIN DD *
IF GROUP TST999 EXISTS
COMMAND DELETE TST999
PROMOTE DLM=ZZ
GROUP (TST999) -
DELETE_EXISTING_TARGET_DATASETS (NO) -
EARLY_DATA_SET_COMPLETION (YES) -
ALLOCSEQ (NONE)
SET -
SPHERE (YES) -
TOLERATE_SOURCE_VALIDATE_FAILURE (YES) -
SOURCE ( DSN( -
DSNCAT.DSNDBD.DBX1ST01.IXA010A.I0001.A001 -
DSNCAT.DSNDBD.DBX1ST01.IXA010B.I0001.A001 -
DSNCAT.DSNDBD.DBX1ST01.IXA010C.I0001.A001 -
DSNCAT.DSNDBD.DBX1ST01.IXA010D.I0001.A001 -
DSNCAT.DSNDBD.DBX1ST01.IXA010E.I0001.A001 -
DSNCAT.DSNDBD.DBX1ST01.IXA010F.I0001.A001 -
DSNCAT.DSNDBD.DBX1ST01.IXA0101.I0001.A001 -
DSNCAT.DSNDBD.DBX1ST01.TSA010A.I0001.A001 ))-
TOLERATE_TARGET_CREATE_FAILURE (YES) -
TARGET (DSN (XXX))
SET -
SPHERE (YES) -
SOURCE ( DSN( -
DSNCAT.DSNDBD.DBX1ST01.TSA010B.I0001.A001 -
DSNCAT.DSNDBD.DBX1ST01.TSA010C.I0001.A001 -
DSNCAT.DSNDBD.DBX1ST01.TSA010D.I0001.A001 -
DSNCAT.DSNDBD.DBX1ST01.TSA010E.I0001.A001 -
DSNCAT.DSNDBD.DBX1ST01.TSA010F.I0001.A001 -
DSNCAT.DSNDBD.DBX1ST01.TSA0101.I0001.A001 -
DSNCAT.DSNDBD.DBX1ST01.TSA0109.I0001.A001 ))-
TARGET (DSN (XXX))
ZZ
COMMAND ACTIVATE TST999
MAXWAIT 180
WAIT GROUP TST999 STATUS DEACT
IF RC = 0
QUIT
Figure 10-11 zDMF Batch JCL Example

10.4.4 Checking Data and Starting DB2 Load
To demonstrate a migration in an active system, Figure 10-12 shows all 15 table spaces (data
consistency check). The record list includes system name, db2 system name, table space,
table name, and a view of the first 10 changed records. The other records were checked as
well. During an error, they are also displayed by the test procedure.
Each record was updated by two different systems (MCECEBC and MZBCVS2). The check
was done from system MCECBC.
:
-- SYSTEM MCECEBC
-- DB2 SYSTEM NAME DB1A
-- DB2 TABLE SPACE TSA010F
-- DB2 TABLE TA010F
-- NUMBER OF RECORDS IN TABLE 25000
-- CHECK ALL RECORDS WHERE FIELD CHANGED1 = YES
-- DISPLAY NUMBER OF UPDATED RECORDS (FIRST 10)
---REC-----SYS------USERID---DATE-----TIME------------UPDATE-MARKED--
0000001 MCECEBC STXXXXX1 20110710 19:53:27.706677 000052 YES
__________ MZBCVS2 STXXXXX2 20110710 19:03:29.100314 000010 YES ___
-- TOTAL RECORDS ----------------------------------------------------
-- MARKED RECORDS = 2500
-- OTHER RECORDS = 0
-- MARKED RECORDS = ALL RECORDS OK
-- OTHER RECORDS = NO PROBLEMS FOUND
*********************************************************************
Figure 10-12 Check data (window shot of last TS result, all others are the same)

Figure 10-13 shows updates from both z/OS systems to these data sets during the migration.
Display Filter View Print Options Search Help

-----------------------------------------------------------------------------
SDSF DA CEBC (ALL) PAG 0 CPU/L 14/ 14 LINE 1-8 (8)
COMMAND INPUT ===> SCROLL ===> CSR
PREFIX=DB* DEST=(ALL) OWNER=* SORT=SIO/D SYSNAME=*
NP JOBNAME StepName ProcStep JobID Owner C Pos DP Real Paging SIO
DB2ADBM1 DB2ADBM1 IEFPROC STC08374 STCUSER NS FE 39T 0.00 4664.4
DB1ADBM1 DB1ADBM1 IEFPROC STC04996 STCUSER NS FE 89T 0.00 3251.2
DB2AMSTR DB2AMSTR IEFPROC STC08372 STCUSER NS FE 2386 0.00 58.06
DB1AIRLM DB1AIRLM STC04994 STCUSER NS FE 19T 0.00 0.00
DB1AMSTR DB1AMSTR IEFPROC STC04992 STCUSER NS FE 2313 0.00 0.00
DB1ADIST DB1ADIST IEFPROC STC04999 STCUSER NS FE 3680 0.00 0.00
DB2AIRLM DB2AIRLM STC08373 STCUSER NS FE 19T 0.00 0.00
DB2ADIST DB2ADIST IEFPROC STC08375 STCUSER NS FE 3715 0.00 0.00
Figure 10-13 SDSF activity display of db2 systems
10.4.5 Starting and Monitoring Test Migration

Submit the BATCH JCL and monitor it using the zDMF Monitor item 1 + 2. The group is
promoted by the BATCH job and automatically start activation. As soon activation is done,
zDMF copies the data sets.
Figure 10-14 shows zDMF Monitor item 1.
Manage Group Members Row 1 to 1 of 1

Primary Commands: CReate EXit Find FILter Switch
Line Commands: C = Copy X = Delete E = Restore M = Modify SMFID CPFX
P = Promote R = Rename V = Verify L = Display CEBC ZD
*******************************************************************************
_ ITSO99 P
*
*
Figure 10-14 zDMF Monitor Item 1 (Manage Groups)
The batch job will end after sending the data to the zDMF master started task.

The JCL log is shown in Figure 10-15.
:
IEF373I STEP/BATCH /START 2011193.0005
IEF032I STEP/BATCH /STOP 2011193.0009
CPU: 0 HR 00 MIN 00.03 SEC SRB: 0 HR 00 MIN 00.00 SEC
VIRT: 240K SYS: 288K EXT: 16468K SYS: 17476K
IEF375I JOB/STRZ999 /START 2011193.0005
IEF033I JOB/STRZ999 /STOP 2011193.0009
CPU: 0 HR 00 MIN 00.03 SEC SRB: 0 HR 00 MIN 00.00 SEC
DWRQBUF§ 2E104120 LN=00FFFEE0 USED=00000000 REQ=00000000
==> IF GROUP ITSO97 EXISTS
==> COMMAND DELETE ITSO99
*** COMMAND SKIPPED DUE TO CONDITION TEST
==> PROMOTE DLM=ZZ

GROUP (ITSO99) -
DELETE_EXISTING_TARGET_DATASETS (NO) -
EARLY_DATA_SET_COMPLETION (YES) -
ALLOCSEQ (NONE)
SET -
SPHERE (YES) -
MIGRATE_ONLY_SPECIFIED_VOLUMES (NO) -
SOURCE ( DSN( -
DSNCAT.DSNDBD.DBX1ST01.IXA010A.I0001.A001 -
DSNCAT.DSNDBD.DBX1ST01.IXA010B.I0001.A001 -
:
)-
) -
TARGET (DSN (XXX))
:
*GROUP PROMOTED, STOW LENGTH IS 2256
*** RC=0
==> COMMAND ACTIVATE ITSO99
GROUP ITSO99 ACTIVATED
*** RC=0
==> MAXWAIT 180
MAXIMUM WAIT SET TO 180
*** RC=0
==> WAIT GROUP ITSO99 STATUS DEACT
MAXIMUM WAIT EXCEEDED
*** RC=4
==> IF RC = 0
==> QUIT
*** COMMAND SKIPPED DUE TO CONDITION TEST
*** MAXIMUM RETURN CODE WAS 4 --> because of DELETE GROUP was skipt
Figure 10-15 zDMF Batch JCL output example

After group is promoted, zDMF checks each data set in the list. It fetches the actual detailed
data set information, number of volumes, multi-volume DS, size, number of extends, and so
on. zDMF also discovers the available space at the new target volumes for each single data
set. After the data sets are validated, the status is changed to Active as shown in
Figure 10-16.
Subsystem ZD Groups Row 1 to 2 of 2

Group CEBC
Data Set Status
Extents
_ ITSO99 Owning System: CEBC Active

Group control blocks initializing (00040/00060)
Figure 10-16 zDMF Monitor Item 2 (Interact with Promoted Groups): Active
As soon all data sets (source and work DS) are Active, zDMF changes to Pending-Active
(Figure 10-17).

Group CEBC
Data Set Status
Extents
_ ITSO99 Owning System: CEBC Moved= 0% P-Active

Activate report available
Figure 10-17 zDMF Monitor Item 2 (Interact with Promoted Groups): Pending Active
The Copy phase starts automatically (Figure 10-18).

Group CEBC
Data Set Status
Extents
_ ITSO99 Owning System: CEBC Moved= 77% Copy

Figure 10-18 zDMF Monitor Item 2 (Interact with Promoted Groups): Copy

When the data is migrated, the status changes to Moved as shown in Figure 10-19.

Group CEBC
Data Set Status
Extents
_ ITSO99 Owning System: CEBC Moved=100% Mirror

Early Data Set Complete status: 000016 Data sets diverted.
Figure 10-19 zDMF Monitor Item 2 (Interact with Promoted Groups): Mirror

After Mirror status is reached, a sequence starts to move non-allocated data sets
(ESDC=YES). Figure 10-20 shows that 16 data sets are already moved from the source data
sets.
DSLIST - Data Sets Matching DSNCAT.DSNDBD.DBX1ST01 Row 1 of 30


-------------------------------------------------------------------------------
DSNCAT.DSNDBD.DBX1ST01.IXA010A.I0001.A001 RS603A+
DSNCAT.DSNDBD.DBX1ST01.IXA010B.I0001.A001 RS6037+
DSNCAT.DSNDBD.DBX1ST01.IXA010C.I0001.A001 RS6031+
DSNCAT.DSNDBD.DBX1ST01.IXA010D.I0001.A001 RS603E+
DSNCAT.DSNDBD.DBX1ST01.IXA010E.I0001.A001 RS6035+
DSNCAT.DSNDBD.DBX1ST01.IXA010F.I0001.A001 RS6030+
DSNCAT.DSNDBD.DBX1ST01.IXA0101.I0001.A001 RS603C+
DSNCAT.DSNDBD.DBX1ST01.IXA0107.I0001.A001 RS603F+
DSNCAT.DSNDBD.DBX1ST01.IXA0108.I0001.A001 RS603A+
DSNCAT.DSNDBD.DBX1ST01.TSA010A.I0001.A001 RS603B+
DSNCAT.DSNDBD.DBX1ST01.TSA010B.I0001.A001 RS6030+
DSNCAT.DSNDBD.DBX1ST01.TSA010C.I0001.A001 RS6032+
DSNCAT.DSNDBD.DBX1ST01.TSA010D.I0001.A001 RS6034+
DSNCAT.DSNDBD.DBX1ST01.TSA010E.I0001.A001 RS6030+
DSNCAT.DSNDBD.DBX1ST01.TSA010F.I0001.A001 RS6032+
DSNCAT.DSNDBD.DBX1ST01.TSA0101.I0001.A001 ZD850C+
DSNCAT.DSNDBD.DBX1ST01.TSA0102.I0001.A001 ZD850E+
DSNCAT.DSNDBD.DBX1ST01.TSA0103.I0001.A001 ZD850F+
DSNCAT.DSNDBD.DBX1ST01.TSA0104.I0001.A001 ZD850F+
DSNCAT.DSNDBD.DBX1ST01.TSA0106.I0001.A001 ZD850D+
DSNCAT.DSNDBD.DBX1ST01.TSA0107.I0001.A001 ZD850B+
Figure 10-20 Data set status (source data sets)

Similarly, Figure 10-21 shows the data sets on the work side. Because the VSAM data sets
are moved, the VSAM CLUSTER name is not displayed. You will find data parts only for
source and work data sets. In the example, eight data sets are already moved, and the VSAM
Cluster has no relation to a volume.
DSLIST - Data Sets Matching XXX.ITSO99.D1193.T0020546.**.DATA Row 1 of 31


-------------------------------------------------------------------------------
XXX.ITSO99.D1193.T0020546.S00001.DATA ZD850F+
XXX.ITSO99.D1193.T0020546.S00002.DATA ZD850B+
XXX.ITSO99.D1193.T0020546.S00003.DATA ZD850C+
XXX.ITSO99.D1193.T0020546.S00005.DATA ZD850D+
XXX.ITSO99.D1193.T0020546.S00009.DATA ZD850E+
XXX.ITSO99.D1193.T0020546.S00022.DATA RS6038+
XXX.ITSO99.D1193.T0020546.S00023.DATA RS603F+
XXX.ITSO99.D1193.T0020546.S00029.DATA RS603A+
Figure 10-21 Data set status (work data sets)

10.4.6 Diverting the Group
Started the divert process by entering V for divert in front of the group as shown in
Figure 10-22.

Group CEBC
Data Set Status
Extents
v ITSO99 Owning System: CEBC Moved=100% Mirror

Early Data Set Complete status: 000016 Data sets diverted.

Group CEBC
Data Set Status
Extents
_ ITSO99 Owning System: CEBC Diverted= 25% P-Divert

Figure 10-22 ZDMF Monitor Item 2 - start divert
To see more details by modifying the display options, press the F4 key and modify the first
three lines (Figure 10-23).
Display Options
Show Source Data Set Name . . . . . . Y or N . . . y <--

Show Target Data Set Name . . . . . . Y or N . . . y <--
Show Extent Information . . . . . . . Y or N . . . y <--
Show Command Messages . . . . . . . . Y or N . . . N
Show Command Diagnostic Information . Y or N . . . N
F1=Help F5=Save Settings F3=Exit

Figure 10-23 ZDMF Monitor Item 2 (modify display options)
Save your new settings by pressing F4.

Pressing F3 shows the new details as shown in Figure 10-24. The details include the number
of all extends for each data set (source and target), and cylinder and head information. VSAM
Cluster information is also displayed.

Group CEBC
Data Set Status
Extents
_ ITSO99 Owning System: CEBC n/a Divert

Activate and Divert reports available
_ Src:DSNCAT.DSNDBC.DBX1ST01.IXA010A.I0001.A001
VSAM Cluster
Tar:XXX.ITSO99.D1193.T0020546.S00001
VSAM Cluster
_ Src:DSNCAT.DSNDBC.DBX1ST01.IXA010B.I0001.A001
VSAM Cluster
Tar:XXX.ITSO99.D1193.T0020546.S00002
VSAM Cluster
:
_ Src:DSNCAT.DSNDBD.DBX1ST01.TSA010F.I0001.A001
Unit=6032(RS6032) Cyl=000004D Hd=B Divert
Unit=6032(RS6032) Cyl=00002D9 Hd=5 Divert
Unit=6032(RS6032) Cyl=00003D7 Hd=5 Divert
Unit=6032(RS6032) Cyl=00004FD Hd=5 Divert
Tar:XXX.ITSO99.D1193.T0020546.S00021.DATA
Unit=850C(ZD850C) Cyl=000015B Hd=1 #Trks=00000976 Moved=100%
Unit=850C(ZD850C) Cyl=000019C Hd=2 #Trks=00000375 Moved=100%
Unit=850C(ZD850C) Cyl=00001B5 Hd=2 #Trks=00000435 Moved=100%
Unit=850C(ZD850C) Cyl=00001D2 Hd=2 #Trks=00000495 Moved=100%
:
_ Src:DSNCAT.DSNDBD.DBX1ST01.TSA0101.I0001.A001
Unit=6038(RS6038) Cyl=000000C Hd=0 Complete
Unit=6038(RS6038) Cyl=000000C Hd=B Complete
Unit=6038(RS6038) Cyl=000038A Hd=0 Complete
Unit=6038(RS6038) Cyl=00008BC Hd=0 Complete
Unit=6038(RS6038) Cyl=0000A35 Hd=0 Complete
Tar:XXX.ITSO99.D1193.T0020546.S00022.DATA
Unit=850C(ZD850C) Cyl=00001F3 Hd=2 #Trks=00000001 Moved=100%
Unit=850C(ZD850C) Cyl=00001F3 Hd=3 #Trks=00000975 Moved=100%
Unit=850C(ZD850C) Cyl=0000234 Hd=3 #Trks=00000375 Moved=100%
Unit=850C(ZD850C) Cyl=000024D Hd=3 #Trks=00000435 Moved=100%
Unit=850C(ZD850C) Cyl=000026A Hd=3 #Trks=00000495 Moved=100%
_ Src:DSNCAT.DSNDBD.DBX1ST01.TSA0102.I0001.A001
Unit=603F(RS603F) Cyl=000000C Hd=0 Complete
Unit=603F(RS603F) Cyl=0000017 Hd=0 Complete
Unit=603F(RS603F) Cyl=000040D Hd=9 Complete
:
Figure 10-24 zDMF Monitor Detailed extend list of an active group

To see which data sets are allocated by application, enter J in front of the group name. The
details on the system SMFID and the application names allocating the data sets are displayed
(Figure 10-25).
Allocated Data Sets for Group ITSO99 Row 1 to 37 of 37

Data Set Name SMFID -------- Job Names -------
DSNCAT.DSNDBC.DBX1ST01.IXA010A.I0001.A001 CVS2 DB2ADBM1
CEBC DB1ADBM1
DSNCAT.DSNDBC.DBX1ST01.IXA010B.I0001.A001 CVS2 DB2ADBM1
CEBC DB1ADBM1
DSNCAT.DSNDBC.DBX1ST01.IXA010C.I0001.A001 CVS2 DB2ADBM1
CEBC DB1ADBM1
DSNCAT.DSNDBC.DBX1ST01.IXA010D.I0001.A001 CVS2 DB2ADBM1
CEBC DB1ADBM1
DSNCAT.DSNDBC.DBX1ST01.IXA010E.I0001.A001 CVS2 DB2ADBM1
CEBC DB1ADBM1
DSNCAT.DSNDBC.DBX1ST01.IXA010F.I0001.A001 CVS2 DB2ADBM1
CEBC DB1ADBM1
DSNCAT.DSNDBC.DBX1ST01.IXA0101.I0001.A001 CVS2 DB2ADBM1
CEBC DB1ADBM1
CEBC DB1ADBM1
CEBC DB1ADBM1
CEBC DB1ADBM1
CEBC DB1ADBM1
CEBC DB1ADBM1
CEBC DB1ADBM1
CEBC DB1ADBM1
CEBC DB1ADBM1
DSNCAT.DSNDBC.DBX1ST01.TSA010A.I0001.A001 CEBC DB1ADBM1
DSNCAT.DSNDBC.DBX1ST01.TSA010B.I0001.A001 CEBC DB1ADBM1
DSNCAT.DSNDBC.DBX1ST01.TSA010C.I0001.A001 CEBC DB1ADBM1
DSNCAT.DSNDBC.DBX1ST01.TSA010D.I0001.A001 CEBC DB1ADBM1
DSNCAT.DSNDBC.DBX1ST01.TSA010E.I0001.A001 CEBC DB1ADBM1
DSNCAT.DSNDBC.DBX1ST01.TSA010F.I0001.A001 CEBC DB1ADBM1
DSNCAT.DSNDBC.DBX1ST02.IXA020C.I0001.A001 CEBC DB1ADBM1
Figure 10-25 zDMF detailed data sets allocation list (zDMF monitor)
Press F3 to go back to the normal monitor display.

In a production move, stay with this status until all data sets in a storage group are moved. In
the test environment, db2 updates were run to other table spaces overnight. The allocation
status changed to the status seen in Figure 10-26.

DSNCAT.DSNDBC.DBX1ST01.IXA010A.I0001.A001 CVS2 DB2ADBM1

DSNCAT.DSNDBC.DBX1ST01.IXA010B.I0001.A001 CVS2 DB2ADBM1
DSNCAT.DSNDBC.DBX1ST01.IXA010C.I0001.A001 CVS2 DB2ADBM1
DSNCAT.DSNDBC.DBX1ST01.IXA010D.I0001.A001 CVS2 DB2ADBM1
DSNCAT.DSNDBC.DBX1ST01.IXA010E.I0001.A001 CVS2 DB2ADBM1
DSNCAT.DSNDBC.DBX1ST01.IXA010F.I0001.A001 CVS2 DB2ADBM1
DSNCAT.DSNDBC.DBX1ST01.TSA010A.I0001.A001 CEBC DB1ADBM1
DSNCAT.DSNDBC.DBX1ST01.TSA010B.I0001.A001 CEBC DB1ADBM1
DSNCAT.DSNDBC.DBX1ST01.TSA010C.I0001.A001 CEBC DB1ADBM1
DSNCAT.DSNDBC.DBX1ST01.TSA010D.I0001.A001 CEBC DB1ADBM1
DSNCAT.DSNDBC.DBX1ST01.TSA010E.I0001.A001 CEBC DB1ADBM1
DSNCAT.DSNDBC.DBX1ST01.TSA010F.I0001.A001 CEBC DB1ADBM1
Figure 10-26 zDMF detailed data sets allocation list (zDMF monitor) next day
In cooperation with the DB2 team, you can complete this group in different ways. You can
close the single table spaces, or you can also stop one DB2 after the other. Wait until all data
sets are in divert status or already moved, and all metadata (Catalog, VTOC and VVDS) are
updated. You can restart of DB2 immediately.
You need to schedule a time, usually on the weekend or during the night, to stop application
for a short time. This outage is required to release allocation to the data sets. As long one
single data set of one session is allocated, the session does not end. As soon application is
accessing the catalog migration of all data sets in status DIVERT, the migration is complete.
Check with zDMF monitor that the group has no data sets allocated as shown in
Figure 10-27.

No jobs allocated to any data sets

included in this group.
Figure 10-27 zDMF Monitor data set allocation status after application was stopped

In addition, make sure that the process is completed as shown in Figure 10-28.

Group CEBC
Data Set Status
Extents
_ ITSO99 Complete
Activate, Divert, and Terminate reports available
Figure 10-28 zDMF Monitor: Status of group is Complete
After group reaches the status Complete, check the new location of the moved and work data
sets as shown in Figure 10-29.
DSLIST - Data Sets Matching DSNCAT.DSNDBD.DBX1ST01.* Row 1 of 30


-------------------------------------------------------------------------------
DSNCAT.DSNDBD.DBX1ST01.IXA0101.I0001.A001 ZD850F+
DSNCAT.DSNDBD.DBX1ST01.IXA0102.I0001.A001 ZD850F+
:
DSNCAT.DSNDBD.DBX1ST01.IXA010E.I0001.A001 ZD850D+
DSNCAT.DSNDBD.DBX1ST01.IXA010F.I0001.A001 ZD850C+
:
DSNCAT.DSNDBD.DBX1ST01.TSA010E.I0001.A001 ZD850D+
DSNCAT.DSNDBD.DBX1ST01.TSA010F.I0001.A001 ZD850C+
DSLIST - Data Sets Matching XXX.ITSO99.D1193.T0020546.**.DATA Row 1 of 30


-------------------------------------------------------------------------------
XXX.ITSO99.D1193.T0020546.S00004.DATA RS603E+
:
Figure 10-29 Location of source and work data sets

Also, test the data consistency by running the test procedure with the check parameter. All
tables were moved to new location without any problem as shown in Figure 10-30.
-- SYSTEM MCECEBC
-- DB2 SYSTEM NAME DB1A
-- DB2 TABLE SPACE TSA010F
-- DB2 TABLE TA010F
-- NUMBER OF RECORDS IN TABLE 25000
-- CHECK ALL RECORDS WHERE FIELD CHANGED1 = YES
-- DISPLAY NUMBER OF UPDATED RECORDS (FIRST 10)
---REC-----SYS------USERID---DATE-----TIME------------UPDATE-MARKED--
-- TOTAL RECORDS ----------------------------------------------------
-- MARKED RECORDS = 2500
-- OTHER RECORDS = 0
-- MARKED RECORDS = ALL RECORDS OK
-- OTHER RECORDS = NO PROBLEMS FOUND
*********************************************************************
Figure 10-30 Test procedure, example result for the last table after migration completed

10.4.7 Starting post migration cleanup procedure
To perform post migration, enter Z in front of the group name as shown in Figure 10-31.

Group CEBC
Data Set Status
Extents
z ITSO99 Complete
Activate, Divert, and Terminate reports available
Figure 10-31 Start Cleanup procedure with line command “Z”
Based on the global setting of the zDMF installation, you can delete the work data sets in one
of two ways:
򐂰 Change the setting for Create ICKDSF TRKFMT Statements to Y as shown in Figure 10-32.
User Session Options

Command ===> 07/17/11 01:25:09
zDMF Command Prefix . . . . . . ZD

Group Definition Data Set . . . STXXXXX.ZDMF.CONFIG
Use Log Data Set . . . . . . . . N (Y/N) Retain 07 generations of logs (01-30)
Use Browse or View . . . . . . V (B/V)
Log Data Set Name . . . . . . ____________________________________________
Messages with Local or GMT time. L (L/G)
Show Command Messages . . . . . N (Y/N)
Show Command Diagnostic Info . . N (Y/N)
Create ICKDSF TRKFMT Statements. Y (Y/N) <-- “Y”
Early Data Set Completion . . . Y (Y/N)
zDMF Load Library . . . . . . . CPAC.HGZD325.GZDLLIB

zDMF Rexx Library . . . . . . . CPAC.HGZD325.GZDELIB
zDMF Panel Library . . . . . . . CPAC.HGZD325.GZDPLIB
zDMF Table Library . . . . . . . CPAC.HGZD325.GZDTLIB
zDMF Message Library . . . . . . CPAC.HGZD325.GZDMLIB
zDMF Security Library . . . . . CPAC.HGZD325.GZDLLIB
F1=Help F3=Exit F5=Save Settings F12 = Cancel

Figure 10-32 zDMF Global setting (zDMF monitor time 6) for ICKDSF TRKFMT
The group TDMF creates a job with ICKDSF track format statements that overwrite all
work data sets with dummy data for each data set migrated. The work data sets and
extend maps created by zDMF are deleted as shown in Figure 10-33 on page 479. You
must update the job card information to your requirements and submit this job.
Remember: This ICKDSF job starts heavy I/O load to the volumes containing the work
data sets.

VIEW SYS11198.T013218.RA000.STXXXXX.R0147420 Columns 00001 00072
****** ***************************** Top of Data *****************************
000001 Jobcard1 Have sufficient region size
000002 Jobcard2 to execute ICKDSF
000003 //*
000004 //* You may submit the jcl from this panel after you
000005 //* update the job card(s). To save a copy, you must use
000006 //* the CREATE or REPLACE primary command to place the
000007 //* data in a data set of your choosing.
000008 //* Reference the ISPF User's Guide on the use of VIEW.
000009 //*
000010 //* All statements should be reviewed prior to submitting.
000011 //*
000013 //* User MUST have authority to execute ICKDSF on volumes.
000014 //*
000015 //* Run ALL the ICKDSF job steps before the data sets are
000016 //* deleted, else the extent information may not be valid.
000017 //*
000018 //* ICKDSF statements for Data Sets successfully migrated.
000019 //*
000020 //S00001 EXEC PGM=ICKDSF
000021 //D00001 DD DISP=OLD,DSN=XXX.ITSO99.D1193.T0020546.S00022
000022 //LDRS6038 DD DISP=OLD,UNIT=3390,VOL=SER=RS6038
000024 //SYSIN DD *
000025
000026 /* Data Set Name: XXX.ITSO99.D1193.T0020546.S00022.DATA */
000027 TRKFMT DDNAME(LDRS6038) VERIFY(RS6038) ERASEDATA CYCLES(3) -
000028 FROMRANGE(00000012,00) TORANGE(00000012,00)
000037 /*
000038 //S00002 EXEC PGM=ICKDSF,COND=(0,NE)
000039 //D00002 DD DISP=OLD,DSN=XXX.ITSO99.D1193.T0020546.S00023
000040 //LDRS603F DD DISP=OLD,UNIT=3390,VOL=SER=RS603F
000042 //SYSIN DD *
000043
000044 /* Data Set Name: XXX.ITSO99.D1193.T0020546.S00023.DATA */
000045 TRKFMT DDNAME(LDRS603F) VERIFY(RS603F) ERASEDATA CYCLES(3) -
000047 TRKFMT DDNAME(LDRS603F) VERIFY(RS603F) ERASEDATA CYCLES(3) -
:
Figure 10-33 Example JCL to start ICKDSF (overwrite work data sets)

Important: Do not delete Extend Maps when zDMF is still active with running groups.
Delete this part of the job, work data sets of completed groups can be deleted.
򐂰 Change the setting for Create ICKDSF TRKFMT Statements to N as shown in Figure 10-34
to create an IDCAMS job with delete statements.
User Session Options

Command ===> 07/17/11 01:25:09
zDMF Command Prefix . . . . . . ZD

Group Definition Data Set . . . STXXXXX.ZDMF.CONFIG
Use Log Data Set . . . . . . . . N (Y/N) Retain 07 generations of logs (01-30)
Use Browse or View . . . . . . V (B/V)
Log Data Set Name . . . . . . ____________________________________________
Messages with Local or GMT time. L (L/G)
Show Command Messages . . . . . N (Y/N)
Show Command Diagnostic Info . . N (Y/N)
Create ICKDSF TRKFMT Statements. N (Y/N) <-- “N”
Early Data Set Completion . . . Y (Y/N)
zDMF Load Library . . . . . . . CPAC.HGZD325.GZDLLIB

zDMF Rexx Library . . . . . . . CPAC.HGZD325.GZDELIB
zDMF Panel Library . . . . . . . CPAC.HGZD325.GZDPLIB
zDMF Table Library . . . . . . . CPAC.HGZD325.GZDTLIB
zDMF Message Library . . . . . . CPAC.HGZD325.GZDMLIB
zDMF Security Library . . . . . CPAC.HGZD325.GZDLLIB
F1=Help F3=Exit F5=Save Settings F12 = Cancel

Figure 10-34 zDMF Global setting (zDMF monitor time 6) for IDCAMS delete statements

Enter Z next to the group to begin the deletion process as shown in Figure 10-35.
VIEW SYS11198.T013954.RA000.STXXXXX.R0147421 Columns 00001 00072

****** ***************************** Top of Data *****************************
000001 Jobcard1 Have sufficient region size
000002 Jobcard2 to execute IDCAMS
000003 //*
000004 //* You may submit the jcl from this panel after you
000005 //* update the job card(s). To save a copy, you must use
000006 //* the CREATE or REPLACE primary command to place the
000007 //* data in a data set of your choosing.
000008 //* Reference the ISPF User's Guide on the use of VIEW.
000009 //*
000010 //* All statements should be reviewed prior to submitting.
000011 //*
000012 //*
000013 //* User MUST have SAF authority to delete these data sets.
000014 //*
000015 //DELETES EXEC PGM=IDCAMS
000017 //SYSIN DD *
000018
000019 /* GROUP NAME: ITSO99 -
000020 DELETE STATEMENTS FOR DATA SETS SUCCESSFULLY MIGRATED. */
000021
000022 DELETE 'XXX.ITSO99.D1193.T0020546.S00022' CLUSTER
:
continued see also next Figure
Figure 10-35 Clean up with data set IDCAMS delete statements for the work data sets

Figure 10-36 shows the results after you start the clean-up procedure.
:
000053
000054 /* GROUP NAME: ITSO99 END */
000055
000056 /*
000057 //EXZD850C EXEC PGM=IDCAMS,DYNAMNBR=1
000059 //SYSIN DD *
000060
000061 /* ATTEMPT TARGET EXTENT MAP DATA SET DELETE ONLY IF */
000062 /* NOT ALLOCATED AND STILL PRESENT ON VOLUME. */
000063 SET LASTCC=0
000064 ALLOCATE -
000065 DSNAME('SYS1.HGZD325.EXTMAP.ZDZD850C.DB') OLD
000066 IF LASTCC EQ 0 -
000067 THEN -
000068 DELETE 'SYS1.HGZD325.EXTMAP.ZDZD850C.DB'
000069 /*
:
Figure 10-36 Clean up with data set IDCAMS delete statements for the Target Extend Map DS
Delete extend map data sets only if you are finished with all sessions. If other sessions are
still active and you want to start cleaning the work data sets, remove the part of the jobs that
deletes the Extend maps.
Do not forget to export the user catalog for the work data sets and delete all aliases for the
work data sets

11
Chapter 11. Using TDMF TCP/IP for z/OS

This chapter describes the TDMF TCP/IP for z/OS product. It explains the additional steps in
relation to a normal TDMF migration required to use the product in a typical remote migration
scenario. The chapter also includes a general guidelines section.

򐂰 TDMF TCP/IP for z/OS overview
򐂰 TDMF general guidelines
򐂰 TDMF TCP/IP example replication

11.1 TDMF TCP/IP for z/OS overview
Transparent Data Migration Facility (TDMF) TCP/IP for z/OS is a host-based software solution
for data replication. It allows migration in a z/OS environment to a new location over a long
distance. TDMF TCP/IP for z/OS is user-initiated and controlled, and guarantees full access
to the local data at any point during a replication operation.
11.1.1 Long-distance data migration software solution

TDMF TCP/IP for z/OS has the following advantages:
򐂰 When the Source or Target Storage System is not IBM storage, or uses different
hardware. Metro Mirror, Global Copy, and other vendor copy services cannot be used or
are not possible.
򐂰 If XRC / GM for z/OS cannot be installed, no HW resources for XRC feature (processor,
cache, HBA, and so on) are available.
򐂰 Easy to manage: No additional automation or management is needed.
򐂰 Easy to install: No need to install additional HW
򐂰 Support any to any storage vendor, hardware independent
11.1.2 Long-distance data migration characteristics

TDMF TCP/IP for z/OS has the following characteristics:
򐂰 It allows long-distance data migration. This type of migration is the copying of data from
one device (the source) to another device (the target) over an Ethernet connection using
TCP/IP.
򐂰 A replication is the logical relationship between a source and target device.
򐂰 The user initiates and controls all migrations. The user identifies the “from” (source)
volumes and the “to” (target) volumes.
򐂰 Multiple volume replication can be established during any single session.
򐂰 Multiple sessions run in parallel, and start sequentially.
򐂰 The replication tool is dynamically activated and manageable.
򐂰 Applications are unaware that replication is underway. The local data is continuously and
fully accessible for read and write activity.
򐂰 For synchronization, all applications must be stopped at a certain point to get consistent
data at the remote site.
򐂰 The tool supports a multiple-system shared data environment.
򐂰 The tool guarantees complete physical data integrity.
򐂰 The use of the tool is not restricted to any control unit model type or device type. Except as
noted, all devices in the data center can participate in replication sessions as required.
򐂰 All volumes (source and target) of a replication must be online.
򐂰 No user can allocate data to a target volume during migration.
Restriction: TDMF for z/OS does not allow access to the target volume during the
migration process.

򐂰 A source volume cannot contain an active local page data set.
򐂰 The source and target volumes must be of the same track geometry.
The proceeding characteristics represent the ideal of a non-disruptive data migration.
11.1.3 Terminology
These terms are used with the TDMF TCP/IP installation and replication process:
Master system (local) The local TDMF system. This system runs as an MVS batch job
that is responsible for the data copy function. Unlike a local TDMF
migration, there are two Master systems in one TDMF session:
Local and remote.
Agent system (local) An associated TDMF MVS image running in a shared local storage
environment with the Master. To ensure data integrity, any MVS
LPAR that has access to the source volumes must be run using
either the Master or an Agent system. The Master and associated
Agent systems communicate through a shared system
communications data set (COMMDS).
Master system (remote) The remote TDMF system. This system runs as an MVS batch job
that is responsible for receiving and storing the data at the remote
site.
Agent systems (remote) Agent systems are allowed at remote site. However, avoid sharing
the remote volumes with other LPARs. TDMF prevents access by
other users and systems to these volumes during the replication. If
the target volumes are shared at remote site, Agents must be
active at these systems.
Source volume DASD volumes containing the data to be replicated.
Target volume DASD volumes receiving the replicated data.
Migration pair The relationship between one source and one target volume for a
single replication.
Migration group A group of volume pairs that use the same group name.
Migration session One Master (local) plus any Agents (local), one Master (remote),
and pairs or groups to be processed in a single TDMF execution.
Data consistency The procedures to ensure that data remains intact and correct
through the migration. All applications at the local site must be
stopped before TDMF TCP/IP can go to synchronization to ensure
consistent data.
Synchronization The time when I/O is inhibited to the source devices (quiesce
phase), all applications stopped. During this time, the last of the
updates are collected and reflected to the target devices
(Synchronization phase). The Volid of the target volume is NOT
changed during this process. It must be changed separately. When
this process is complete, the TDMF session ends and the target
device is marked offline (depends on replication parameter).
Chapter 11. Using TDMF TCP/IP for z/OS 485

11.1.4 TDMF z/OS components architecture
Figure 11-1 depicts the TDMF z/OS architecture and components. In this example, the
volumes are being moved from a local to a remote site. All storage vendors are supported
(any to any).
The following components are depicted:

򐂰 TDMF TSO monitor: The TDMF TSO Monitor is used to manage the active replication jobs
and view active or past replications. It is installed at the same time as the TDMF tool. The
TDMF TSO Monitor Feature consists of REXX execs, which require ISPF version 3.6 and
TSO/E Version 2.4 at minimum.
򐂰 Communications data set (local / remote): The local communication data set (COMMDS)
is used to pass information between local systems participating in a TDMF TCP/IP
session. This data set contains the status and messages related to a specific session. The
COMMDS also serves as the input file to the TDMF TSO monitor. A COMMDS is also
needed at the remote site, and need to have the same size (number of cylinders).
TDMF
Monitor
(TSO User)
TCP/IP Link
Local Remote Master
Master TDMF
TDMF
Local
Agent
TDMF
ESCON/FICON ESCON/FICON
Director Director
Remote
Copy over IP
Replicate
over TCPIP
COMMDS link COMMDS
Storage Storage
(any vendor) (any vendor)
Local
Remote
Figure 11-1 TDMF TCP/IP Basic overview
򐂰 Master and Agents: Each LPAR that has access to the storage subsystem and shares the
source volumes must run an Agent. Each Master/Agent session must have access to its
own communication data set. Multiple Agent systems can be involved in a session locally.
The remote site normally has only a single Master and no Agents.
Important: To avoid a possible data integrity exposure, all systems accessing replicated
volumes must be identified to the Master system. TDMF includes various controls and
checks that ensure that the user does not make the following errors:
򐂰 Assign or direct conflicting migrations to the same devices
򐂰 Attempt migrations to nonexistent devices
򐂰 Use the same COMMDS for two simultaneous or overlapping migration sessions

Master system responsibilities
The TDMF Master system is responsible for the following tasks:
򐂰 Initialize the TDMF Master system environment and the COMMDS.
򐂰 Start and control each session for all participating systems.
򐂰 Check the communication with the remote site (remote master).
򐂰 Monitor source volume user I/O activity to detect updates.
򐂰 Copy data from the source volume to the target volume through TCP/IP.
򐂰 Process detected source volume updates from all systems.
򐂰 Perform refresh operations to the target site to reflect the update activity on the source
volumes.
򐂰 Check the internal health of the Master environment (both sites) and the health of all
Agent systems.
򐂰 The remote Master system receives data from local site and writes that data to the target
volumes. It also monitors target volume system I/O activity to prevent updates from others
to the target volumes.
Agent system responsibilities

The TDMF Agent system is responsible for the following items:
򐂰 Initialize the TDMF Agent environment and establish communications to the Master
system through the COMMDS
򐂰 Acknowledge and process replication requests (volumes to be copied) from the Master
system
򐂰 Monitor source volume user I/O activity and detect updates
򐂰 Notify the Master system of source volume update activity through the COMMDS
򐂰 Check the internal health of the Agent environments and the health of the Master system
11.1.5 TDMF z/OS hardware compatibility

TDMF is designed to support the count key data/extended (CKD/E)-capable control units of
any vendors as a source or a target storage subsystem. This functionality includes
Hyper-Volume, Flexible-Volumes, and all user-defined volume sizes. Extended address
volumes (EAVs) might be supported depending on the z/OS version. For more information,
see the TDMF Installation and reference manual.
Because TDMF is a host software migration tool, there are no particular hardware
prerequisites.
11.1.6 Install TDMF TCP/IP z/OS

For the installation procedure, see Chapter 9, “Using TDMF for z/OS” on page 395.
You need to install TDMF at the remote site, but you do not need any license key to run TDMF
TCP/IP.

11.1.7 TDMF z/OS customer requirements
TDMF TCP/IP for z supports all MVS-based operating systems currently supported by IBM. It
is not suitable for use with native z/VM and VM/ESA, native VM/XA, native VSE/ESA, native
VSE/XA, or DOS. The Operating Systems Support Matrix is at:
http://www-950.ibm.com/services/dms/en/support/tdmf/zos/matrix.html
Tip: Perform periodic checks of the Required IBM Maintenance and Technical Information
Bulletins (TIBs). These requirements must be implemented to ensure successful TDMF
operation.
Required IBM Maintenance:

http://www-950.ibm.com/services/dms/en/support/tdmf/zos/apars.pdf
IBM Technical Support:

http://www-950.ibm.com/services/dms/en/support/tdmf/zos/v5/ptf530.html
Technical Information Bulletins:

11.2 TDMF general guidelines

This section offers general guidelines when using TDMF TCP/IP z/OS.
11.2.1 Keeping current

It is important to make sure before any migration that your TDMF TCP/IP z/OS code is current
and up-to-date.
򐂰 Bookmark the TDMF Technical Support website. It contains information to aid in the
successful execution of TDMF TCP/IP.
򐂰 Register for automatic email notification. Registration ensures that you will be notified
whenever the Technical Support website is updated. You will receive the actual link for
registration with your license key (by mail).
򐂰 Get a Group ID (Lotus Notes). Click Notification Registry on the Technical Support
website.
򐂰 Download each PTF as it becomes available. When a new PTF is released, the download
links to previous PTFs are removed. The list fixes, however, continue to be available. PTFs
can be installed at a later time.
򐂰 For a non-SMP/E installation, TDMF PTFs are not available and must be installed. The
license key must be saved before installing.
򐂰 Manuals are periodically updated. To ensure that you have the most current information,
check for new Technical Information Bulletins (TIBs). TIBs are typically issued between
manual releases.
򐂰 Make sure that your TSO user ID is allowed to start TDMF TCP/IP and ICKDSF.
򐂰 Create a volume list. The target volumes must have the same number of cylinders as the
source volume or more. Make sure that you do not have duplicates (VOLID, source, or
target device) in your list. Mark your system-related volumes in this list (IPL, IODF, XCF
DS, Local pages, PLPA & COMMON DS, and so on) so they are easy to find at the remote
site.

򐂰 Group volumes in relation to application data (storage groups) into the same session. You
might need to group volumes by control units because of performance issues.
Performance problems might be caused by the ESCON connection or HW with less
internal bandwidth. Make sure that same volume is not defined in different groups.
򐂰 The TCP/IP link speed between the two sites depends on the following factors:
– Number of TB to be replicated
– The time available for production move
– The need to shut down applications at the local site
– IPL at the remote site.
For more information, see 11.3.4, “TDMF TCP/IP performance example” on page 512.
򐂰 Consult with staff in charge of the client network people to establish a range of port
numbers that can be used for replication. The number of ports needed depends on total
number of sessions. However, remember that one TDMF session needs two ports in
sequence. Make sure that ports are not blocked by firewalls between the two sites.
򐂰 Reserve a number of volumes for local communication data sets (COMMDS). Allocate no
more than 3-4 COMMDS for each volumes to prevent performance issues when all
sessions are running. The total number of COMMDS must be the same as number of
sessions. Allocate the same number of COMMDS at the remote site.
򐂰 Define available and active initiators (one initiator for each session) for both sites. The
service ratio must be SYSSTC or similar for all TDMF jobs
򐂰 Create local temporary Local Page data sets and do a page delete to the original local
page data sets. TDMF TCP/IP does not copy active local page data sets. The temporary
local page data sets are not copied to the remote site.
򐂰 Establish naming conventions for the job names because all jobs must run in parallel at all
LPARs. Naming conventions can help avoid confusion. For more information, see 11.3.2,
“Naming conventions” on page 497.
򐂰 Schedule the migration for a time when heavy system I/O load such as database
reorganization and BATCH work can be stopped.
Normally replication of storage volumes is combined with a tape/library system replication. It

is also sometimes done in combination with open storage systems and servers. Therefore, it
is important to talk to the whole team responsible for the replication. This publication
addresses only the disk part.
11.2.2 Creating a consistent data structure

To move an entire data center, you must have a consistent data structure. This section
highlights the components of a consistent data structure. At the local site, you have one or
more processors, and one or more storage systems. At the remote site, you have a minimum
of one z/OS system to receive the data.
The data structure has the following components. Figure 11-2 on page 490 shows how these
components relate to each other.
򐂰 Local system: The local system consists of all volumes in relation to the system. Examples
include IPL, IODF, local page data sets, and system-related volumes.
򐂰 TDMF volumes: These volumes are TDMF libraries and COMMDS volumes. These
volumes are not moved.

򐂰 Temporary local page data sets: These data sets are on volumes that are not replicated.
TDMF cannot copy active local page data sets, so the original data sets are removed from
the page pool by a page delete command. They can then be replicated.
򐂰 Data volumes: This term includes all volumes, including storage management system
(SMS) and non-SMS managed volumes with applications, application data, and temporary
work data sets.
򐂰 Remote system: This system includes all volumes at the remote site in relation to the
remote master system. This system receives the data from local master over IP and writes
the data to the target volumes in relation to the defined volume pairs.
򐂰 TDMF Monitor: This TDMF REXX ISPF application runs at a TSO user ID. This user must
have access to all LPARs running TDMF Master/Agents sessions.
򐂰 TCP/IP Link: This link is the physical connection from local to remote site using TCP/IP as
the protocol. The normal bandwidth for a service engagement (2-12 TB) is a 1 GBit link
speed. With this connection, you can expect up to120 MBps at the link.
TDMF
Monitor
(TSO User)
TCP/IP Link
Local Remote Master

TDMF
Master TDMF
Local
Agent
TDMF
Storage Storage
Local Remote
Temp
System
: : System
= Temp TDMF
TDMF
local page DS
:
System
: (Remote)
Data : : Data
Replication, Volid:
:
remote copy over IP : XXdddd
: before
migration
starts
Figure 11-2 TDMF TCP/IP overview for a data center move to a new location
11.2.3 Summary of replication tasks and steps

Before you begin the replication, perform the following steps:
򐂰 Do system assurance for both sites, including the TCP/IP connection. For more
information, see the TDMF Installation and Reference Manual.
򐂰 Check that all volumes are available and online on the target system (new storage).
򐂰 Install TDMF TCP/IP on both sites with the systems programmer.
򐂰 Define the volume pairs (source / targets) with the storage administrator.

򐂰 Create TDMF JOBs for both sites.
򐂰 Create ICKDSF INIT / Rename Jobs for the remote site.
򐂰 Prepare temporary local page data sets on the volumes that are not moved. These data
sets are necessary because active local pages cannot be handled by TDMF.
򐂰 Do a first network test to check the link, connection, and bandwidth.
The replication is normally done in three stages. Along with the storage administrator, update
the volume list (relation source and target volumes) is updated to the actual status before
each s5age.
1. Perform a test without synchronization data as shown in Figure 11-3. This test is normally
done with system/network volumes only, and is also a network test. Application data are
copied to test the bandwidth.
TDMF TCPIP – Sequence of Replication
1. Test Move
Non consistent Data
Must be done because of
Init all Rename all
Target volumes volumes to
setting the correct VOLID to
their original name the volumes and update
vary them online
different hardware at remote
site to the VTOC
1 Remote master
Start
2 Local Agents (if needed) Note all changes must be
sessions Check, modify
3 Local master made to the system
System for first IPL
parameter for the new
Check Check with customer location (IODF, sysparm., . . )
TCPIP bandwidth and performance
connection
IPL at remote site
for test
Data consistency Test IPL and network
When initial
copy is finished, is not important for a first
go to sync IPL/Network test When test is finished
shutdown system at remote site
Repeat this test if needed
Local site Remote site
Figure 11-3 Test without sync data

2. Run a test with the complete synchronization data to test the system, network, and
application data at remote site (Figure 11-4).
2. Test Move
With consistent data
Initialize all Rename all Must be done because of
Target volumes volumes to setting the correct VOLID to
their original name the volumes and update
vary them online different hardware at remote
site to the VTOC
Remote master
Start
Local Agents (if needed)
sessions
Local master
Clients responsibility
Check Check with customer
Check, modify Note all changes must be
TCPIP bandwidth and performance
System for IPL made to the system
connection
parameter for the new
location (IODF, sysparm., . . )
When initial
copy is finished, Stop all applications, bring Test Network and applications
keep refresh mode IPL at remote site
down Systems except VTAM,
for a test
TCPIP, TSO, RACF and
TDMF sessions.
Go to sync Go to sync (manual When test is finished
command)
shutdown system
Restart systems at
local site Repeat this test if needed

Figure 11-4 Test with synchronization data, Production Test Environment (PTE)

3. After completing the tests, run the actual production move (Figure 11-5). All data must be
copied again because no updates were performed during the tests. In addition, there was
no relation between the local and remote sites.

Production Move
Initialize all With consistent data Rename all
Must be done because of
Target volumes volum es to setting the correct VOLID to
vary them online their original nam e the volumes and update
different hardware at remote
site to the VTOC
Remote master
Start
sessions
Local Agents (if needed)
Local master Clients responsibility
Check, modify
Check Check with customer System for IPL DO all changes to the system
TCPIP performance bandwidth an performance parameter for the new
connection location (IODF, sysparm., . . )
IPL Systems
When initial remote site
copy is finished,
keep refresh mode Stop all applications, bring
down system except VTAM, Part of customer:
TCPIP, TSO, RACF and Check Network
connection
Check all data
TDMF sessions.
Go to sync (internal/external) Define a point of no return !
Go to sync Check all application
Shutdown systems GO
at local site

Figure 11-5 Production move, sequence of replication

11.2.4 Link performance information
Figure 11-6 shows the performance found in test environments for various connection types,
and the expected throughput for TDMF TCP/IP.
Connection Bytes Per MB Per MB per Hour GB per TDMF

Type Minute Minute Hour Speed
GB per
Hour
384K DSL 2,949,120 2.8125 168.75 .1647 .1318
T1 11,520,000 10.98 658.18 .6437 .5149
10 base T 75,000,000 71.52 4,291.53 4.19 3.35
Ethernet
T3 345,600,000 329.589 19,775 19.31 15.45
100 Base T 750,000,000 715.255 42,915 41.90 33.52
Ethernet
OC3 1,162,500,000 1,108.64 66,518 65.95 51.96
OC12 4,500,000,000 4,291.53 257,492 251.45 201.16
Gigabit 7,500,000,000 7,152.55 429,153 419.09 335.27
Ethernet
OC48 18,000,000,000 17,166.13 1,029,968 1,005.82 804.66
OC192 72,000,000,000 68,664.52 4,119,872 4,023.28 3,218.64
Figure 11-6 Link speed examples

Figure 11-7 shows performance information for an example involving moving 7 TB (1424
volumes) over 400 km with a dedicated 1 GBit link. The information was collected every 15
minutes from the IBM RMF™ monitor at the local and remote sites. The sessions were not
started in sequence. Information from the test move showed that the best results came from
putting the volumes/sessions with high update rates at the end of the migration windows.
Model Cyl count
3390-1 1113 128

3390-3 3339 576
3390-9 10017 560

3390-27 30051 160
Total 1424
Figure 11-7 List with RMFMON performance information

Figure 11-8 shows the impact on sessions running with software compression ON or OFF for
the same example. With software compression ON, the link operated at not more than 120
MBps. However, the FICON channels at the remote site can achieve up to 160 MBps. In the
example, software compression cannot be used during online business because of the high
processor load. This setting cannot be changed dynamically.
7TB, 400 km distance, dedicated OSA adapter, 1 Gbit link
software compression active
Figure 11-8 Example: Performance information client production move (TDMF TCP/IP 5.2)
11.3 TDMF TCP/IP example replication

This example shows how to create TDMF, ICKDSF Rename, and ICKDSF INIT jobs. It
includes an example of naming conventions. The network test is also included with screen
captures of the JCL, monitor, and some performance information.
The distance in this example was around 100 km between local and remote site. The client
wanted to test the connectivity and bandwidth for the two new defined links of 1 GBit each. To
retain the performance information, the test was run twice and the resulting information was in
different COMMDS. Sessions 01 - 03 were in the first test, and sessions 04 - 06 were in the
second.
Both tests used the same 30 volumes (10 out of each DISK Storage unit). Therefore, after the
first test the RENAME and INIT jobs were run, and all 30 volumes were set ONLINE again. All
devices with 2xxx and 3xxx addresses were attached with ESCON Channel paths. The rest
were FICON attached. Example 11-1 shows the list of volumes there were used.
Example 11-1 Volume list for test

Source volumes at source site
UC2010 - UC2019 (ESCON connection)
UC3010 - UC3019 (ESCON connection)
UC4010 - UC4019 (FICON connection)
Target volumes at remote site
UC8400 – UC8409 (FICON connection)

UC8500 – UC8509 (FICON connection)
UC8600 - UC8609 (FICON connection)
LPAR SYCP ran the master session at local site. All others ran agents. LPAR SYX1 ran the
master session at remote site. There were no agents at the remote site. The target volumes
were not shared.
11.3.1 Overview of replication

Figure 11-9 shows a simplified view of the physical setup of the example. The Channel Path
IDs (CHP) are listed to check the performance using the performance monitor (RMFMON) of
z/OS.
Clients 192.168.2.12
z/OS environment Router Router Remote
192.168.2.35 site
Local Site
CHP: 28 CHP: 7A
IP connection setup
CHP: 01
Network test and CHP: 2A
Production Test
Environment (PTE)
(2 * 1Gbit)
2086.A04.SN: 38xx x 2096.S07.SN: 36xxx
TDMF
TSO
Monitor
SY06 (remote)
SY07 SYCD
SYFP SYCP (M)
SYFD
LPARs SYCR SYX1 (M)
SYX2
TSO: 10.27.190.123 TSO: 10.28.64.73
SY02
DISK
DISK DISK DISK
2000 - 23FF 3000 - 33FF 4000 - 47FF

CHP: 34 44 5E 70 CHP: 0A 1D 71 84 5000 - 5BFF 7000 – 7FFF
2400 – 27FF 3400 – 37FF CHP: 05 2E 2F 30 55 56 57 7E CHP: 2C 2D 2E 2F 34 35 36 37
CHP: 10 1E 45 63 CHP: 3C 49 62 73 8000 – 8AFF
CHP: 1C 1D 1E 1F 30 31 32 33
CHPIDs connected to SYCP
Figure 11-9 TDMF TCP/IP test replication overview
11.3.2 Naming conventions

Before creating the JCLs, create naming conventions to avoid confusion during the migration
process. The following is an example from a service project.
These sessions require one master and six agents for each session. Therefore, there are
seven LAPARs. Two tests are run with three sessions each to test different links. All of the
information is stored in the COMMDS.
In total, 42 jobs (3 (sessions) * 7 (# of LPARs) * 2 (tests)) need to be created.

The naming conventions used are shown in Example 11-2.
Example 11-2 Job naming convention (example)

//TDMLCP01 JOB .....
!!! !
!!! +--> session number (2 digits)
!!+----> 2 char out of the LPAR name (first or last 2 char)
!! to identify AGENTs at different LPARs, make sure not have
!! duplicate jobnames, only the first one will start.
!+-----> L=Local site, R=Remote site
+------> M=Master, A=AGENT

JCL examples at remote site
The job names generated at the local site are shown in Figure 11-10 on page 499.
EDIT STXXXXX.TDMF.NWTS1.TDMF.CNTL Row 00001 of 00044

_________ $OVR00 --> Overview
_________ TDALCD01 Agents for SYCD
_________ TDALCD02 “
_________ TDALCD03 “
_________ TDALCD04 “
_________ TDALCD05 “
_________ TDALCD06 “
_________ TDALCR01 Agents for SYCR
_________ TDALCR02 “
_________ TDALCR03 “
_________ TDALCR04 “
_________ TDALCR05 “
_________ TDALCR06 “
_________ TDALFD01 Agents for SYFD
_________ TDALFD02 “
_________ TDALFD03 “
_________ TDALFD04 “
_________ TDALFD05 “
_________ TDALFD06 “
_________ TDALFP01 Agents for SYFD
_________ TDALFP02 “
_________ TDALFP03 “
_________ TDALFP04 “
_________ TDALFP05 “
_________ TDALFP06 “
_________ TDAL0601 Agents for SY06
_________ TDAL0602 “
_________ TDAL0603 “
_________ TDAL0604 “
_________ TDAL0605 “
_________ TDAL0606 “
_________ TDAL0701 Agents for SY07
_________ TDAL0702 “
_________ TDAL0703 “
_________ TDAL0704 “
_________ TDAL0705 “
_________ TDAL0706 “
_________ TDCL$G00 <-- create COMMDS
_________ TDMLCP01 Master for SYCP
_________ TDMLCP02 “
_________ TDMLCP03 “
_________ TDMLCP04 “
_________ TDMLCP05 “
_________ TDMLCP06 “
**End**
Figure 11-10 Job names in our test replication (local site)

In a large environment, it can be hard to find the jobs that belong together. However, since
z/OS 1.8 there is a new function, build in ISPF, to set a filter for a partitioned data set
extended (PDSE). This filter function is shown in Figure 11-11.
DSLIST - Data Sets Matching MBSM.TDMF.NWT* Row 4 of 4


-------------------------------------------------------------------------------
e /(*01) MBSM.TDMF.NWTS2.TDMF.CNTL TDM001
Figure 11-11 Filter function for PDSE members
The result of this filter command is shown in Figure 11-12. All members for session 01 are
displayed, making it is easy to submit the right jobs.
EDIT MBSM.TDMF.NWTS1.TDMF.CNTL Row 00001 of 00007

_________ TDALCD01
_________ TDALCR01
_________ TDALFD01
_________ TDALFP01
_________ TDAL0601
_________ TDAL0701
_________ TDMLCP01
**End**
Figure 11-12 Display of PDSE members fit to the filter criteria (*01) only

For long-distance data migration, you must specify a new parameter for replication to address
the remote master. Specify an IP address and a port number as shown on line 22 in
Figure 11-13.
EDIT MBSM.TDMF.NWTS1.TDMF.CNTL(TDMLCP01) - 01.00 Columns 00001 00072

****** ***************************** Top of Data *****************************
000001 //TDMLCP01 JOB (,DASD),'IBM DATA MIGRATION',CLASS=T,
000002 // REGION=0M,MSGCLASS=X,NOTIFY=&SYSUID
000003 //* TDMF DATA MIGRATION - DO NOT CANCEL THIS JOB !!!
000004 //* telefon number IBM service person 1
000006 /*XEQ TXRPRD
000007 /*JOBPARM SYSAFF=SYCP
000008 //******************************************************
000009 //* Master JOB TDMF running: SYCP local site
000010 //******************************************************
000011 //LOCAL01 EXEC PGM=GTDMAIN,PARM=MASTER,REGION=0M
000012 //STEPLIB DD DISP=SHR,DSN=SYS1.GTD530.GTDLLIB
000013 //GTDKEY DD DISP=SHR,DSN=SYS1.GTD530.GTDLLIB
000014 //SYSCOM DD DISP=SHR,DSN=SYS1.TDMF.SDNT.LOCAL.SYSCOM01
000017 //SYSIN DD *
000018 SESSION M(DG72)
000019 AGENTS( DH72 DG72 DI72 DK72 GA98 GB98)
000020 OPT(FAST UNIDENT(W) PROMPT CONC(4 ACTIVE)
000021 NOCOMPRESS PAC CHECKT OFFLINET SINGLE)
000022 REMOTE DX72 ADDR(192.168.2.12) PORT(8200)
000023 REPLICATE UC2010 UC8400 OPT(MAXTR(1))
****** **************************** Bottom of Data ***************************
Figure 11-13 TDMF master JCL local site (TDMLCP01)

In Figure 11-14 shows an example for a local Agent job.
EDIT MBSM.TDMF.NWTS1.TDMF.CNTL(TDALCD01) - 01.00 Columns 00001 00072

****** ***************************** Top of Data *****************************
000001 //TDALCD01 JOB (,DASD),'IBM DATA MIGRATION',CLASS=T,
000006 /*XEQ TORCSD
000007 /*JOBPARM SYSAFF=SYCD
000008 //******************************************************
000009 //* AGENT JOB TDMF running: SYCD
000010 //******************************************************
000011 //STEP01 EXEC PGM=GTDMAIN,PARM=AGENT,REGION=0M
000014 //SYSCOM DD DISP=SHR,DSN=SYS1.TDMF.SDNT.LOCAL.SYSCOM01
****** **************************** Bottom of Data ***************************
Figure 11-14 TDMF Agent JCL local site (TDALCD01)
The following are other parameters used in this example:

FAST Copy allocated space only.
UNIDENT(W) Check for other systems without an agent, and send a warning if one
is found. For a test or production move, use UNIDENT(T) so if an
unknown system is available, the session terminates.
PROMPT Stop at REFRESH status and wait for confirmation before going to
sync.
CONC(4 ACTIVE) Run four volumes concurrently. This value can be changed
dynamically later using the TDMF monitor.
NOCOMPRESS Do not use software compression.
PAC Use pacing to automatically reduce the number of tracks per I/O
during a performance problem.
CHECKT Check the target volume for data, and stop the migration if data are
allocated there.
OFFLINET After session ends, set the target volumes offline.
SINGLE Combines all volumes into a single session that you can end (go to
sync) using a manual command.
MAXTRACKS(15) This parameter adjusts the number of tracks read per I/O and the
memory usage in the common area. If you run out of memory, use
MAXTR(1) to reduce the performance impact. This parameter is also
useful for old devices attached using ESCON.
For more information about the parameters, see the TDMF Installation and Reference
Manual.

JCL examples at remote site
The remote site requires an additional JCL. The migration then proceeds with the following
steps:
1. The TDMF master jobs receive the data and write it to the target volumes. The list used in
the example is shown in Figure 11-15.
DSLIST - Data Sets Matching MBSM.TDMF.NWTS1 Row 1 of 3


------------------------------------------------------------------------------
MBSM.TDMF.NWTS1.INIT.CNTL TDM001
MBSM.TDMF.NWTS1.RENAME.CNTL TDM001
MBSM.TDMF.NWTS1.TDMF.CNTL TDM001
Figure 11-15 TDMF and ICKDSF PDSE list used for Network test
The remote site is shown in Figure 11-16.
EDIT MBSM.TDMF.NWTS1.TDMF.CNTL Row 00001 of 00009

_________ $OVR00 <-- Overview
_________ TDCR$G00 <-- create COMMDS
_________ TDMRX101 Master for SYX1
_________ TDMRX102 “
_________ TDMRX103 “
_________ TDMRX104 “
_________ TDMRX105 “
_________ TDMRX106 “
_________ VDASIN01 <-- device list (UCBs)
**End**
Figure 11-16 TDMF Master jobs remote site
Tip: Do not define the source or target volumes at the remote site. The port number in
relation to the local session port number is important. The remote master listens to the
TCP/IP port defined in the JCL. As soon the TCP/IP session is established, this port is
used to exchange the volume information. The following port number (port 8201) is
used at the remote site to establish for each source/target volume pair an own TCP/IP
session. See Figure 11-34 on page 513.

The Master JCL remote site is shown in Figure 11-17.
EDIT MBSM.TDMF.NWTS1.TDMF.CNTL(TDMR0601) - 01.00 Columns 00001 00072

****** ***************************** Top of Data *****************************
000001 //TDMRX101 JOB (,DASD),'IBM DATA MIGRATION',CLASS=T,
000006 //*XEQ SYX1
000007 /*JOBPARM SYSAFF=SYX1
000008 //******************************************************
000009 //* MASTER JOB TDMF SYX1 remote site
000010 //******************************************************
000011 //STEP01 EXEC PGM=GTDMAIN,PARM='MASTER,PORT=8200'
000014 //SYSCOM DD DISP=SHR,DSN=SYS1.TDMF.SDNT.REMOTE.SYSCOM01
000018 //SYSIN DD *
000019 SESSION M(DX72)
000020 OPT(FAST UNIDENT(T))
****** **************************** Bottom of Data ***************************
Figure 11-17 TDMF Master JCL remote site
2. After replication, rename the target volumes to their original VOLID. Otherwise, the client
cannot work with these volumes for test. To rename the volumes, use ICKDSF
REFORMAT, which is a z/OS system tool (Figure 11-18).
EDIT MBSM.TDMF.NWTS1.RENAME.CNTL Row 00001 of 00006

_________ TDRR$G01
_________ TDRR$G02
_________ TDRR$G03
_________ TDRR$G04
_________ TDRR$G05
_________ TDRR$G06
**End**
Figure 11-18 ICKDSF RENAME jobs

The results of the rename job on session 01 are shown in Figure 11-19.
EDIT MBSM.TDMF.NWTS1.RENAME.CNTL(TDRR$G01) - 01.0 Columns 00001 00072

****** ***************************** Top of Data *****************************
000001 //TDRR$G01 JOB (,DASD),'IBM DATA MIGRATION',CLASS=T,
000006 //STEP0001 EXEC PGM=ICKDSF,REGION=3000K,PARM='NOREPLYU'
000007 //SYSPRINT DD SYSOUT=(*,,1TAB)
000008 //SYSIN DD *
000009 REFORMAT UNIT(8400) VFY(UC8400) VOLID(UC2010)
****** **************************** Bottom of Data ***************************
Figure 11-19 ICKDSF RENAME job example for target volumes in session “01”
3. The ICKDSF init jobs initialize the target volumes back to their old VOLID. Therefore, you
do not have to change the TDMF JCL again (Figure 11-20).
EDIT MBSM.TDMF.NWTS1.INIT.CNTL Row 00001 of 00006

_________ TDMI$G01
_________ TDMI$G02
_________ TDMI$G03
_________ TDMI$G04
_________ TDMI$G05
_________ TDMI$G06
**End**
Figure 11-20 ICKDSF INIT jobs

The results of the init jobs on the target volumes is session 01 are shown in Figure 11-21.
EDIT MBSM.TDMF.NWTS1.INIT.CNTL(TDMI$G01) - 01.00 Columns 00001 00072

****** ***************************** Top of Data *****************************
000001 //TDMI$G01 JOB (,DASD),'IBM DATA MIGRATION',CLASS=T,
000006 //STEP0001 EXEC PGM=ICKDSF,REGION=3000K,PARM='NOREPLYU'
000007 //SYSPRINT DD SYSOUT=(*,,1TAB)
000008 //SYSIN DD *
000009 INIT UNIT(8400) VFY(UC2010) VOLID(UC8400)
****** **************************** Bottom of Data ***************************
Figure 11-21 ICKDSF INIT job example for target volumes in TDMF session “01”
Rename the volumes to their original VOLIDs after replication so you can start tests. Before
the next test or the production move, you must switch it back. Use the ICKDSF verify (VFY)
parameter in combination with NOREPLY=U, so you can run this job without a manual REPLY
at the operator console.

11.3.3 TDMF TCP/IP examples for a network test
The following shows the first session and what is going on in the TDMF Monitor and the RMF
performance monitor.
1. Start the local Master and local Agents as shown in Figure 11-22.


ComDataSet : SYS1.TDMF.SDNT.LOCAL.SYSCOM01
SINGLE UC2010 Inactive
Figure 11-22 TDMF Monitor Item 1 (local session started)
TDMF is waiting now for the remote master. As soon the TDMF remote master is started,
TDMF starts the copy process for the defined concurrent number of volumes.
After a few seconds, you see the estimated copy phase end time behind each volume as


SINGLE UC2010 Copy -------------------> 0:01:12
SINGLE UC2011 Copy -------------------> 0:01:11
SINGLE UC2012 Copy --------------------> 0:01:05
SINGLE UC2013 Copy -------------------> 0:01:13
Figure 11-23 TDMF Monitor, start copy phase

2. If there is no impact to the production, increase the concurrent number of volumes by
switching to TDMF Monitor Item 2. Dynamically change the number of volumes running
concurrently for example to 8 as shown in Figure 11-24.
Session Status Row 1 to 10 of 10


ComDataSet SYS1.TDMF.NWTS1.LOCAL.SYSCOM01 Sessions 01
__ UC2010 2010 SINGLE Copy PIT P 005
UC8400 192.168.002.012 Port=08200

__ UC2011 2011 SINGLE Copy PIT P 005
UC8401 192.168.002.012 Port=08200

:
Figure 11-24 TDMF menu item 2, change number of concurrent volumes
3. Switch back to TDMF menu item 1 pressing F10. There are now four additional running
volumes, for a total of eight (Figure 11-25).


SINGLE UC2010 Copy --------------------------> 0:01:24
SINGLE UC2011 Copy --------------------------> 0:01:23
SINGLE UC2012 Copy ---------------------------> 0:01:17
SINGLE UC2013 Copy --------------------------> 0:01:25
SINGLE UC2014 Copy -----> 0:05:04
SINGLE UC2015 Copy -----> 0:04:58
SINGLE UC2016 Copy -----> 0:05:07
SINGLE UC2017 Copy -----> 0:05:03
Figure 11-25 TDMF menu item 1

After a few minutes, you see the window shown in Figure 11-26.


SINGLE UC2010 Refresh 1 ------------------------------------------------>
SINGLE UC2014 Copy ------------------------------> 0:01:59
SINGLE UC2015 Copy ------------------------------> 0:02:01
SINGLE UC2016 Copy -------------------------------> 0:01:57
SINGLE UC2017 Copy -------------------------------> 0:01:58
SINGLE UC2018 Copy ---> 0:04:33
SINGLE UC2019 Copy ---> 0:04:40
Figure 11-26 TDMF menu item 1
4. When the initial copy phase is done, you see the updates (here Refresh x) and TDMF
refreshes the volumes asynchronously. The message SYNC Volume Needed means that this
session is ready to go to SYNC (Figure 11-27).
Load is running to the source volumes
Session Monitor SYNC Volume Needed.


SINGLE UC2015 Refresh 2 --------------------------->
Figure 11-27 TDMF menu item 1, session in refresh phase

In this example, the updates to the source volumes were stopped to emulate an
application stop. Start the SYNC phase using the PG command in front of the first volume
(Figure 11-28). You must confirm the command at the request window.
Session Status SYNC Volume Needed.


ComDataSet SYS1.TDMF.SDNT.LOCAL.SYSCOM01 Sessions 01
pg *UC2010 2010 SINGLE Waiting Sync PIT P 005
UC8400 192.168.002.012 Port=08200

__ UC2011 2011 SINGLE Refresh 1 PIT P 005
UC8401 192.168.002.012 Port=08200

UC8402 192.168.002.012 Port=08200

UC8403 192.168.002.012 Port=08200

Figure 11-28 TDMF Monitor menu item 2, stop the session, go to SYNC, line command “PG”

TDMF starts the SYNC phase as shown in Figure 11-29. The upper left area displays the
total numbers of volumes in this session. Below this number, a counter displays the
increasing number of completed volumes.

ComDataSet SYS1.TDMF.SDNT.LOCAL.SYSCOM01 Sessions 01
__ UC2010 2010 SINGLE Synchronize PIT P 005
UC8400 192.168.002.012 Port=08200

__ UC2011 2011 SINGLE Complete PIT P 005
UC8401 192.168.002.012 Port=08200

__ UC2012 2012 SINGLE Complete PIT P 005
UC8402 192.168.002.012 Port=08200

__ UC2013 2013 SINGLE Synchronize PIT P 005
UC8403 192.168.002.012 Port=08200

:
Figure 11-29 TDMF Monitor menu item 2, start SYNC phase
As soon all volumes within the session are completed, the session is no longer displayed
in the TDMF monitor. The jobs belonging to this session end. Check all job log to see
whether MAXCC is greater than zero (Figure 11-30).
06.21.19 JOB02611 $HASP165 TDMRX101 ENDED AT TXRTEST MAXCC=0 CN(INTERNAL)

Figure 11-30 TDMF job info at remote site
For the local master job MAXCC was equal to four as shown in Figure 11-31. All warning
messages result in a MAXCC=4. In this test, there was an additional LPAR with access to
the replicated source volumes. No Agent was started at this LPAR because the source
volumes were OFFLINE.
06.21.14 JOB02609 $HASP165 TDML0601 ENDED AT TXRPROD MAXCC=0 CN(INTERNAL)

06.21.16 JOB02612 $HASP165 TDML0701 ENDED AT TXRPROD MAXCC=0 CN(INTERNAL)
06.21.17 JOB02616 $HASP165 TDMLFD01 ENDED AT TXRPROD MAXCC=0 CN(INTERNAL)
06.21.18 JOB02611 $HASP165 TDMLCR01 ENDED AT TXRPROD MAXCC=0 CN(INTERNAL)
06.21.18 JOB02618 $HASP165 TDMLFP01 ENDED AT TXRPROD MAXCC=0 CN(INTERNAL)
06.21.17 JOB02619 $HASP165 TDMLCD01 ENDED AT TXRPROD MAXCC=0 CN(INTERNAL)
06.21.20 JOB02607 $HASP165 TDMLCP01 ENDED AT TXRPROD MAXCC=4 CN(INTERNAL)
Figure 11-31 TDMF job info at local site
If MAXCC is greater than four, check the job logs for error messages. You will need to
restart the session.

11.3.4 TDMF TCP/IP performance example
In this test, RMFMON was run and displayed the channel load for the Open System adapter
(OSA) CHPID 28 (Figure 11-32).
There was not more than 64 MBps at the OSA connection. Because the source volumes were
ESCON attached, there might not be more than 15 M/sec each channel path. There were four
paths to the control unit plus TCP/IP usage.
F CPU= 11/ 11 UIC= 65K PR= 0 SYCP CHANNEL T

06:07:15 CHANNEL UTILIZATION(%) READ(B/S) WRITE(B/S) FICON OPS ZHPF OPS
ID NO G TYPE S PART TOT BUS PART TOT PART TOT RATE ACTV RATE ACTV
28 OSD Y 46.5 47.2 35.6 227K 229K 64M 64M
29 OSE Y 0.0 0.3 7.3 0 0 0 0
Figure 11-32 RMFMON Channel statistics
11.3.5 TCP/IP information

This section covers information about the TCP/IP connections
TCP/IP settings for TDMF replication

The following TCP/IP settings must be checked and updated for test/production move. The
TCP/IP send and receive buffers should have these values listed in Figure 11-33, which are
256 KB each.
You can view this information using the TSO netstat config command.
EZZ2350I MVS TCP/IP NETSTAT CS V1R11 TCPIP Name: TCPIPBC 04:52:02

EZZ2700I Home address list:
EZZ2701I Address Link Flg
EZZ2702I ------- ---- ---
EZZ2703I 192.168.2.04 LOSAEG09 P
EZZ2703I 127.0.0.1 LOOPBACK
EZZ2350I MVS TCP/IP NETSTAT CS V1R11 TCPIP Name: TCPIPBC 04:52:02
EZZ2720I TCP Configuration Table:
EZZ2721I DefaultRcvBufSize: 00262144 DefaultSndBufSize: 00262144
EZZ2727I DefltMaxRcvBufSize: 00262144 SoMaxConn: 0000000010
EZZ2722I MaxReTransmitTime: 120.000 MinReTransmitTime: 0.500
EZZ2723I RoundTripGain: 0.125 VarianceGain: 0.250
EZZ2724I VarianceMultiplier: 2.000 MaxSegLifeTime: 30.000
EZZ2725I DefaultKeepALive: 00000120 DelayAck: No
EZZ2726I RestrictLowPort: Yes SendGarbage: No
EZZ2728I TcpTimeStamp: Yes FinWait2Time: 300
EZZ2729I TTLS: No
:
Figure 11-33 Result of TCP/IP netstat config command
TCP/IP port mapping for a TDMF session

For each TDMF TCP/IP session, you need two ports at the remote site. The defined port in
the TDMF Master JOBs (local and remote site) is used to establish access to the other site. It
is also used for communication between the two master jobs.

In this example, the port numbers shown in Figure 11-34 are used. Each volume pair (source
/ target volume) has a TCP/IP session. At the local site, TCP/IP is using a free port mapped to
the port 8201 for each volume pair. In the TDMF example, there are 10 volume pairs in the
TDMF session. Therefore, you see 10 different ports in relation to port 8201 at the remote site
transferring the data. There is also one port in relation to port 8200 at the remote site for the
communication between the two master sessions.
To see these TCP/IP sessions, use the TSO netstat conn command.
EZZ2350I MVS TCP/IP NETSTAT CS V1R11 TCPIP Name: TCPIPCP 04:52:02

EZZ2585I User Id Conn Local Socket Foreign Socket State
EZZ2586I ------- ---- ------------ -------------- -----
EZZ2587I BPXOINIT 00000013 0.0.0.0..10007 0.0.0.0..0 Listen
EZZ2587I DB1ADIST 0000349D 0.0.0.0..5051 0.0.0.0..0 Listen
EZZ2587I DB1ADIST 0000349E 0.0.0.0..5054 0.0.0.0..0 Listen
EZZ2587I DB1ADIST 000034A2 0.0.0.0..5052 0.0.0.0..0 Listen
EZZ2587I FTPDB1 00013FE7 0.0.0.0..21 0.0.0.0..0 Listen
:
EZZ2587I TDMLBC01 000587A7 192.168.2.04..2230 192.168.2.12..8201 Establ
sh
sh
EZZ2587I TDMLBC01 000587AF 192.168.2.04..2238 192.168.2.12..8201 Establ
sh
EZZ2587I TDMLBC01 000587AE 192.168.2.04..2237 192.168.2.12..8201 Establ
sh
sh
EZZ2587I TDMLBC01 000587AC 192.168.2.04..2235 192.168.2.12..8201 Establ
sh
EZZ2587I TDMLBC01 000587AA 192.168.2.04..2233 192.168.2.12..8201 Establ
sh
EZZ2587I TDMLBC01 00058787 192.168.2.04..2198 192.168.2.12..8200 Establ
sh
EZZ2587I TDMLBC01 000587AD 192.168.2.04..2236 192.168.2.12..8201 Establ
sh
EZZ2587I TDMLBC01 000587AB 192.168.2.04..2234 192.168.2.12..8201 Establ
sh
sh
:
Figure 11-34 Viewing the TCP/IP session port mapping for a TDMF session

A
Appendix A. Network block devices for Linux

This appendix addresses techniques where block devices are presented to a host, but are
physically attached to another host. This configuration allows you to migrate data in a
host-based manner for Linux operating systems.
This appendix contains the following sections:

򐂰 Network block devices
򐂰 Summary

Network block devices
Network block devices are block devices where the real device is attached to a different Linux
host. The device accepts the same block commands to write, read, and maintain data as a
conventional disk. However, the block device forwards the I/Os through a network using TCP
or UDP protocols to a server where the actual device is connected to
This methodology allows you to migrate data per blocks to a different destination. Depending
on the implementation, migrations can use low-level copy commands like dd or cpio. Some
implementations allow an RAID-1 function where the data is mirrored automatically to two
different network block devices.
The Linux Network Block Device ndb is a basic implementation of a generic network block
device. It has been included in the kernel since version 2.1.101. It provides read-only and
read/write access to a remote device. Many users mount the remote device as read only. In
this configuration, only one host is allowed write access the remote device. However, there is
no access control included in the code. This configuration carries the risk that data can be
deleted if more than one client tries to write to this network device.
When the ndb server is started, it listens to a TPC or UDP port assigned by the administrator.
The server is started with the command shown in Example A-1.
Example A-1 Starting the ndb-server

# ndb-server 5000 /dev/sda1
# _
In this example, the ndb device is represented by a file in the directory /export. When the
server starts, it listens to port 5000 for incoming requests.
After the server is started, you can create a block device by referencing the IP address of the
server and the port where the server listens (Example A-2).
Example A-2 Create a ndb device

# ndb-client 9.155.33.67 5000 /dev/ndb0
# _
The network block device can be read to create a file system, mount it, and allocate files and
directories. For data migrations, however, you need to use other methods.
Suggested methods
Creating a file system in the network block device and copying the data file by file is a basic
migration technique. A more efficient method is to copy the whole device in a single dd
statement to the network device as shown in Example A-3.
Example A-3 Copying a raw device to the nbd

# dd if=/dev/sda2 of=/dev/nbd0 bs=1024
# _
Using method means that the applications using the data on the disk must be shut down. In
addition, the file system must be unmounted to capture all recent I/Os from the cache.
Depending on the used capacity of the disk and the bandwidth of the network, this process
might take a while.

In the following sections, other processes that reduce the downtime because the
synchronization can be done in the background are addressed. With these methods, the
applications only need to be stopped for the cut over to the new storage.
Using software RAID1

The easiest way to replicate the data to the network device is to establish an RAID1 as a
software RAID using the tool mdadm. The mdadm tool can provide different RAID levels using
real block devices. Data migration using RAID1 involves the following steps:
1. Implement RAID1, which is mirroring between two volumes. After it is created, both
volumes are synchronized in the background.
2. When the synchronization is completed, all updates by applications or users are replicated
to both devices.
3. The migration is finalized when the file system is unmounted and you close RAID1 using
the mdadm tool.
The following steps show a simple and straightforward example:

1. Close all applications and unmount the file system.
2. When the RAID1 device is established, a new device is allocated that writes the data to
both devices. Mount the file system to this device as shown in Example A-4.
Example A-4 Create the RAID1 out of the local disk and the nbd0 device
# mdadm --create /dev/md0 --level=mirror --raid-devices=2 /dev/sda2 /dev/nbd0
mdadm: /dev/sda2 appears to contain an ext2fs file system
size=1048576K mtime=Thu Jul 7 14:30:34 2011
Continue creating array? y
mdadm: array /dev/md0 started.
#
3. The device /dev/sda2 is the source which must be transmitted to the network block device
/dev/nbd0. The result is a new device /dev/md0, which is now the RAID1 device.
4. The synchronization of both mirrors starts immediately. The progress of the
synchronization can be monitored with the command shown in Example A-5.
Example A-5 Query the status of the synchronization

# cat /proc/mdstat
1048512 blocks [2/2] [UU]
[======>..............] resync = 31.2% (327680/1048512) finish=0.5min
speed=23405K/sec

# _
When the synchronization is complete, you see the status shown in Example A-6.
Example A-6 Status, when synchronization is done

# cat /proc/mdstat
Appendix A. Network block devices for Linux 517

1048512 blocks [2/2] [UU]

# _
5. Mount the file system for the applications after RAID1 is established (Example A-7).
Example A-7 Mount the RAID1 device

# mount /dev/md0 /mnt
# df
Filesystem 1K-blocks Used Available Use% Mounted on
/dev/sda1 14761044 8922208 5089004 64% /
none 4071508 412 4071096 1% /dev
none 4076096 1932 4074164 1% /dev/shm
none 4076096 188 4075908 1% /var/run
none 4076096 0 4076096 0% /var/lock
none 4076096 0 4076096 0% /lib/init/rw
/dev/md0 1032088 103468 876192 11% /mnt <===
# _
6. Any updates from users or the applications are mirrored to both underlaying devices.
When you are ready for the cut over, close the applications and unmount the file systems,
and then close RAID1 as shown in Example A-8.
Example A-8 Cut over to the target device

# umount /mnt
#
# mdadm --stop /dev/md0
mdadm: stopped /dev/md0
#
# cat /proc/mdstat
# _
7. When RAID1 is established, a md superblock is written to each device. To mount the

remote device on the target system, remove this superblock as shown in Example A-9.
Example A-9 Clean up the superblock at the target system

# mdadm --zero-superblock /dev/nbd0
#
# mount /dev/sda1 /APPLICATION
# _
8. The migration is now complete. Stop the nbd-server at the target system, stop the
nbd-client at the source system, and clean up RAID1.
Summary
The Network Block Device nbd can be used for basic migration to new Linux systems and
storage devices at different sites. This migration technique is effective for migrations involving
only a few disks.

However, two major issues must be considered first:
򐂰 The communication between the server and the client is realized by specifying a TCP user
port. Using a port might be a security issue in some environments.
򐂰 There is no control mechanism that prevents more than one client from writing data to the
server. Having no control mechanism creates a high risk of corrupting the data on the
server. You must ensure that no other client has access to the server device.
Appendix A. Network block devices for Linux 519

B
Appendix B. CLI Conversion

This chapter describes how to migrate from the ESS Copy Services tasks environment to the
DS Copy Services environment. The Copy Services interfaces described are the ESS
graphical user interface (GUI) and the command-line interface (CLI). The Copy Services
interfaces that are run directly by z/OS are not included.
This appendix includes the following sections:

򐂰 Migrating ESS Copy Services tasks to DS8000 CLI
򐂰 Copy Services commands

Migrating ESS Copy Services tasks to DS8000 CLI
All of the Advanced Copy Services functions available on the ESS are available on the
DS8000. However, there are differences that need to be considered when replacing the ESS
CLI with the DS8000 CLI:
򐂰 Point-in-Time Copy (FlashCopy) does not support Consistency Groups in the DS Storage
Manager.
򐂰 Fibre Channel is used for Metro Mirror, Global Copy, Global Mirror, and Metro/Global
Mirror.
򐂰 The DS Storage Manager runs real time only (no saved tasks), whereas the DS8000 CLI
can be started real time or with a saved script.
򐂰 The DS8000 CLI supports both the ESS 800/750 and the DS8000.
򐂰 ESS must be at Licensed Internal Code (LIC) level 2.4.3.15 or higher.
Review the ESS tasks to migrate

Review what ESS Copy Services tasks you want to migrate. The tasks can be viewed from
the ESS GUI or the ESS CLI. Figure B-1 shows the saved tasks listed in the ESS GUI.
Figure B-1 ESS Copy Services GUI tasks panel
Highlight the task and click the information panel. Clicking the panel displays the task
information that will be used for converting to the DS8000 CLI. The information from the task
Flash10041005 is shown in Figure B-2. This task establishes a FlashCopy relationship
between source volume 004-23953 and target volume 005-23953 with the No Background
Copy option.
Figure B-2 ESS task information

The same task list can be generated using the ESS CLI command esscli list task as
shown in Figure B-3. This output shows the same tasks as shown with the GUI.
esscli list task –s copy_services-server –u csadmin –p passw0rd
Wed Nov 24 10:29:31 EST 2004 IBM ESSCLI 2.4.0

Task Name Type Status
------------------------------------------------------------------------
H_Epath_test16 PPRCEstablishPaths NotRunning
H_Epath_test17 PPRCEstablishPaths NotRunning
Brocade_pr_lss10 PPRCEstablishPair NotRunning
Brocade_pr_lss11 PPRCEstablishPair NotRunning
Flash10041005 FCEstablish NotRunning
Figure B-3 The esscli list task
Select Flash10041005 and use the ESS CLI command esscli show task to look at the details
of the task (Figure B-4). In this output, the source and target volumes are listed as 1004 and
1005. This LUN format is used in the DS8000 CLI.
esscli show task –s copy_services_server –u csadmin –p passw0rd –d

“name=Flash10041005”
Wed Nov 24 10:37:17 EST 2004 IBM ESSCLI 2.4.0

Taskname=Flash10041005
Tasktype=FCEstablish
Options=NoBackgroundCopy
SourceServer=2105.23953
TargetServer=2105.23953
SourceVol TargetVol
------------------------------------------------------------------------
1004 1005
Figure B-4 The esscli show task
Review the specific server scripts that perform tasks that set up and start saved ESS CLI
tasks. These server scripts might have to be edited or translated to run the equivalent
DS8000 CLI commands.
Appendix B. CLI Conversion 523

Another example of a saved task list with the ESS GUI is displayed in Figure B-5. In this
example, there are four saved tasks to establish and remove PPRC paths and pairs.
Figure B-5 PPRC task list
In the task to establish PPRC paths named tstpths, there is a single path in the list from
source 21968:00 to 21968:01 (Figure B-6).
Figure B-6 Establish PPRC paths task

Look at the task named est1011 to establish the PPRC pairs in the information panel as
shown in Figure B-7. This example is establishing Extended Distance pairs (now known as
Global Copy). The information panel displays 11 volumes in source and target columns.
Figure B-7 Establish PPRC pairs task

The next saved task in the list is named del01. This task terminates the pairs created in
est1011 (Figure B-8).
Figure B-8 Terminate PPRC pairs information window
Important: Open systems volume IDs on the ESS are given in an 8-digit format:
xxx-sssss, where xxx is the LUN ID and sssss is the serial number of the ESS. The fixed
block LUN ID used for DS8000 CLI must be the xxx number in the ESS format plus 1000. If
you start the flash copy command on the ESS with the DS8000 CLI, the fixed block LUNs
must be specified as 1xxx to avoid overwriting the following CKD volumes:
򐂰 0000 to 0FFF: System z CKD volumes (4096 possible addresses)
򐂰 1000 to 1FFF: Open systems fixed block LUNs (4096 possible addresses)
Convert the individual tasks

This section list the ESS CLI tasks that will be translated to equivalent DS8000 CLI
commands. These commands can be saved as scripts to be run in the DS8000 CLI
environment. The commands can also be run as real-time commands using DS8000 CLI in
interactive mode.
Important: Unlike the ESS GUI, the DS8000 DS Storage Manager does not save Copy
Services tasks. The DS Storage Manager is only used in real-time mode for Copy Services
functions.

Table B-1 lists the conversions of task parameters from the ESS CLI format to the DS8000
CLI format for the FlashCopy task reviewed in Figure B-4 on page 523.
Table B-1 ESS CLI to DS8000 CLI conversion

Task parameter ESS CLI parameter DS8000 CLI Description
conversion
Tasktype FCEstablish mkflash Establish FlashCopy
Options NoBackgroundCopy -nocp No background copy

when establishing the
FlashCopy relationship
SourceServer 2105.23953 -dev IBM.2105-23953 Storage unit device ID
TargetServer 2105.23953 N/A DS8000 CLI does not

use a target server for
FlashCopy commands
Source and Target 1004 1005 1004:1005 Source volume

volumes separated by a colon
from the target volume
in DS8000 CLI
Table B-2 lists the conversions of ESS Copy Services parameters from the DS8000 CLI
format for the PPRC task est1011. This table focuses on establishing PPRC as related to the
saved task rather than all the saved tasks shown previously. The remaining tasks are
converted to DS8000 CLI in the next section.
Table B-2 ESS Copy Services to DS8000 CLI conversion

Task parameter ESS Copy Services DS8000 CLI Description
parameters conversion
Tasktype Establish PPRC mkpprc Establish extended

Extended Distance distance PPRC
(Global Copy)
Options Do Not Copy Vol -nocp Establish pairs with no

Secondary Online OK -tgtonline background copy and
PPRC Extended -type gcp target volume online
Distance as Global Copy
SourceServer 2105-21968 -dev IBM.2105-21968 Source storage unit

device ID
TargetServer 2105-21968 -remotedev Remote storage unit

IBM.2105-21968 device ID
Source and Target 0080 1080 0080-008B:1080-108B Source and target

volumes ... volume ranges
separated by a (:)
colon in DS8000 CLI

Use the information in Table B-1 on page 527 to convert the ESS CLI commands to establish
FlashCopy into DS8000 CLI commands. Issue the mkflash command using the No
Background Copy option -nocp to the ESS volumes 1004:1005 (Figure B-9). The relationship
is queried using the DS8000 CLI command lsflash for the specific pair.
dscli> mkflash -nocp -dev IBM.2105-23953 1004:1005

CMUC00137I mkflash: FlashCopy pair 1004:1005 successfully created.
dscli> lsflash -dev IBM.2105-23953 1004:1005

ID SrcLSS SequenceNum Timeout ActiveCopy Recoding Persistent Revertible
===============================================================================
1004:1005 10 0 120 Disabled Disabled Disabled Disabled
Figure B-9 DS8000 CLI command to establish and list a FlashCopy pair
To create the PPRC paths seen in Figure B-6 on page 524, use the DS8000 CLI command
mkpprcpath (Figure B-10). This example shows a single path between two ESS storage units
using source LSS 00 and target LSS 01. This output is slightly different than in the information
panel for tstpths because that task is creating paths within the same ESS.
dscli> mkpprcpath -dev IBM.2105-23953 -remotedev IBM.2105-21968 -remotewwnn

5005076300C09368 -srclss 00 -tgtlss 01 I0004:I0084
Figure B-10 Creating PPRC paths on ESS with DS8000 CLI
To create the PPRC pairs seen in Figure B-7 on page 525, use the DS8000 CLI command
mkpprc as shown in Figure B-11. Use the following options:
򐂰 The -type gcp option to create the extended distance PPRC relationship
򐂰 The -mode nocp option to specify the Do Not Copy Vol parameter
򐂰 The -tgtonline option to specify the Secondary Online OK parameter
Select the source volume range 0080 to 008B paired with target volumes 0180 to 018B, as
shown in the ESS task est1011.
dscli> mkpprc -dev IBM.2105-23953 -remotedev IBM.2105-21968 -type gcp -mode

nocp -tgtonline 0080-008B:0180-018B
Figure B-11 Create PPRC pairs on ESS with DS8000 CLI
The final task is converting the task to terminate PPRC, which is displayed in Figure B-8 on
page 526, using the DS8000 CLI command rmpprc. This conversion is shown in Figure B-12.
The saved task on the ESS did not include any options for the termination. Therefore, the new
DS8000 CLI command does not include any of the available options for this command.
dscli> rmpprc -dev IBM.2105-23953 -remotedev IBM.2105-21968 0080-008B:0180-018B

Figure B-12 Terminate PPRC pairs on ESS with DS8000 CLI

Copy Services commands
The Flash Copy and RMC Copy Services DS8000 CLI commands and the equivalent ESS
Copy Services commands are listed in Table B-3. The commands are grouped by type of
function. The ESS CLI can start saved tasks created through the ESS GUI by issuing the
command rsExecuteTask. The functions that are executed in a saved task have a
corresponding DS8000 CLI command. This table is the correlation between the functions
started in a saved task and the DS8000 CLI command.
Table B-3 DS8000 CLI copy services commands and equivalent ESS CLI commands
DS8000 CLI command ESS Copy Services Description
Flash Copy Commands
mkflash FCEstablish Establishes FlashCopy
lsflash rsFlashCopyQuery Lists FlashCopy relationship
mkremoteflash FCEstablish with Inband option Creates FlashCopy pair on

remote system through RMC
conduit
resyncflash FCEstablish with Increment Updates a point in time copy,

and Persistent options copying only the tracks that
have been updated since the
last copy
resyncremoteflash FCEstablish with Inband, Same as resyncflash except on

Increment, and Persistent a remote system issued
options through an RMC conduit
reverseflash FCEstablish with Reverse Reverses the FlashCopy

Restore option direction so that the source
becomes the target
revertremoteflash FCEstablish with Inband and Restores data on the source

Revertible options volume to its most recent
consistency formation, issued
through RMC conduit
rmflash FCWithdraw Removes the FlashCopy

relationship
rmremoteflash FCWithraw with Inband option Removes the FlashCopy

relationship on the remote
storage unit through RMC
conduit
unfreezeflash Flash Copy create consistency Resets a FlashCopy

group consistency group previously
established with the -freeze
parameter
PPRC Commands
failbackpprc PPRCEstablishPair with PPRC Copies data from the source

Failback option volume to the target volume to
resume mirroring
failoverpprc PPRCEstablishPair with PPRC Site switch causing the target

Failover option volumes to become the source
volumes

DS8000 CLI command ESS Copy Services Description
freezepprc PPRCFreezeGroup Creates new RMC consistency

group and removes RMC paths
lsavailpprcport n/a Lists ports available for RMC

paths between two storage
units
lspprc n/a Displays the RMC status

information
lspprcpath list pprcpaths Displays RMC paths for the

specified source and target
LSSs
mkpprc PPRCEstablishPair Establishes RMC pairs
mkpprcpath PPRCEstablishPaths Establishes RMC paths
pausepprc PPRCSuspendPair Pauses an existing RMC

relationship
resumepprc n/a Resumes RMC relationship
rmpprc PPRCTerminatePair Removes RMC relationship
rmpprcpath PPRCRemovePaths Removes RMC paths
Copy Services notes

Consider the following notes about Copy Services:
򐂰 A 2107 ESCON I/O port is only used for System z host attachment. It cannot be
configured as an RMC path.
򐂰 A 2107 Fibre Channel port is configured for either SCSI-FCP or FICON protocol. Like the
2105, a FICON port is restricted to the point-to-point/switched fabric topology setting. A
FICON I/O port is used for System z host attachment and cannot be configured as an
RMC path.
򐂰 A 2107 Fibre Channel SCSI-FCP I/O port can be configured for the point-to-point
(switched) fabric of FC-AL connection topologies. A port that uses point-to-point
(switched) fabric topology can be used for Open systems host I/O and RMC path
configurations simultaneously.
򐂰 A 2107 storage facility image can contain up to 65,280 volumes. A 2105 storage unit can
contain up to 4096 FB volumes and 4096 CKD volumes.

Related publications
The publications listed in this section are considered particularly suitable for a more detailed
discussion of the topics covered in this book.
IBM Redbooks
For information about ordering these publications, see “Help from IBM” on page 531. Note
that some of the documents referenced here may be available in softcopy only.
򐂰 IBM System Storage DS8000: Copy Services in Open Environments, SG24-6788
You can search for, view, download or order these documents and other Redbooks,
Redpapers, Web Docs, draft and additional materials, at the following website:
ibm.com/redbooks
Other publications
These publications are also relevant as further information sources:
򐂰 DS8000 Command-Line Interface User’s Guide, SC26-7625-05
򐂰 IBM System Storage DS8000 User’s Guide, SC26-7915
򐂰 z/OS DFSMS Advanced Copy Services, SC35-0428
Online resources
These websites are also relevant as further information sources:
򐂰 Data Migration Service
http://www.ibm.com/servers/storage/datamobility/
򐂰 IBM System Storage
http://www.ibm.com/systems/storage/index.html
Help from IBM

IBM Support and downloads
ibm.com/support

IBM Global Services
ibm.com/services

Index
cluster 148–150
A CNTL 412
access method services (AMS) 450 command-line interface (CLI) 47
ACF2 448 COMMDS see communications data set (COMMDS)
activation phase 401 communication links 454
Active in Copy 422 communications data set (COMMDS) 396, 398, 401,
active/active 135, 137, 143, 168, 173 403, 412, 419, 485–487
137 completion phase 435
active/passive 135, 137, 139, 168, 173 concurrent instances 199
Add Mirror 279 concurrent volumes 429
agent system 396, 485 configuration node 178
responsibilities 398, 487 consistency groups 182, 522
AIX LVM 330 consolidation 266
alias 436 Copy Pending 87
ALLOCCM 419 copy phase 401, 435, 446
ALLOWmirrorchange 400 Copy Services tasks 526
ALTER authority 399 Couple data sets 424
alternate couple data set 424 coupling facility resource management (CFRM) 456
alternate path 327 policy 456
AMS see access method services (AMS) CQUERY 74
authentication 51 cquery 100–101
auxiliary SVC cluster 246 create pairs 71–72
auxiliary volume 183 create paths 67
cross-system coupling facility (XCF) 454, 456
B CS Domain 54
back-end application 178
background copy 135–136, 148, 158 D
backup 10 DASDVOL 405
bandwidth 44 Data Facility Storage Management Subsystem (DFSMS)
basic catalog structure (BCS) 440, 450 437
basic disk 268 data integrity 398, 420, 458, 486
BCS see basic catalog structure (BCS) data migration 133–135, 142
blockdev 382 back-out 148, 169–170
checklist 170
C monitoring 158
cache 43, 91, 162, 431 object 146–148, 157
destage 186 offline 185
catalog structure 450 online 188
CCHH 425 steps 138, 150, 169
cell type 451 synchronization 151
CESTPAIR 71, 75 target 135
cfgmgr 90, 334, 343, 357 volume 146
CFRM see coupling facility resource management data redirection 437
(CFRM) Data Set Services (DFSMSdss) 438
channel 425 datapath 277
extension 55 dataset level migration 434
CHECK TARGET 421 DB2 444
checkpoint 434 dd 9–10
chfs 92 deadlock 447
chunks 185–186 deadly embrace 458
CICS 444 delete pairs 72
CKD 49 dense wave division multiplexing (DWDM) 43
CLASS=DATASET 405 dependent writes 182
CLI see command-line interface (CLI) Destination Name 146, 157

Destination Pool 146, 157 FlashCopy 176
device geometry 12 freezepprc 60
Device Manager 161 front-end application 178
device mapper 376 fsck 92
device special file 322, 329 Full Duplex 72, 87
df command 92, 292, 299, 308 Full Speed Copy 422, 425, 429
DFDSS 444 fuzzy copy 45
DFSMdss see Data Set Services (DFSMSdss)
DFSMS see Data Facility Storage Management Subsys-
tem (DFSMS) G
DFSMSdfp 438 GDG see generation data group (GDG)
DFSMShsm see Hierarchical Storage Manager (DFSM- generation data group (GDG) 420, 439
Shsm) Global Copy performance 46
DIAGNOSE 451 global enqueue 458
disk group 319 Global Mirror 42, 185
Disk Magic 46 global resource serialization (GRS) 454, 457
diskpart 164 go to sync 45, 73
disruptive 297 graphical user interface (GUI) 47
diversion 447, 457 GRS see global resource serialization (GRS)
Diversion phase 435, 457 GUI see graphical user interface (GUI)
divert 447
dm_list 150, 153 H
DS CLI 49, 522 Hardware Management Console (HMC) 47
DS Storage Manager 47 hardware reserve 456–457
DS4000 138, 144 HBA 267, 289, 305
DS6000 137 firmware level 267
DS8000 137 hdisk 337
firmware 267 Hierarchical Storage Manager (DFSMShsm) 438
serial number 334 History option 399, 404
dual cluster Metro Mirroring 188 HMC see Hardware Management Console (HMC)
DWDM see dense wave division multiplexing (DWDM) host attachment procedures 150
dynamic disk 268, 272 host computer 176
Dynamic Multipathing (DMP) 307 host server 134–135, 150
dynamic swap 434 I/O requests 135
dynamic takeover 396 HP-UX Volume Manager 320
E I
ENQ 458 I/O enclosure 55
environment variables 155–156 I/O group 178
ESCON 49, 56, 425 IBM XIV
ESS 137, 144, 157, 330, 350 data migration solution 134
CLI 522 storage 135
GUI 522 ICF catalog 450
port 322 ICKDSF 47, 50, 72, 425, 443
ESS Specialist 52 IDCAMS 436, 449
ESTPAIR 72 idling 259
event log 160 IEBCOPY 403
EXAMINE 453 image mode MDisk 194
Extend Volume 286 image mode VDisk 185, 196
Extended Distance 525 importvg 90–91
Extended Mirrored phase 441 Inhibit Cache Load 431
extent 179, 181, 186 initialization phase 400–401
initialization rate 159
F initiator 134, 137, 139
fail-over 137, 168 Input/Output Supervisor (IOS) 430
FastCopy 422, 425 insf 322
FCEstablish 527 Intercluster Metro Mirror 184
fdisk 381 intercluster Metro Mirror 184, 188
FICON 49, 425 Intracluster Metro Mirror 184

intracluster Metro Mirror 184, 188 VDisk 251
IOCDS 99 master system 396, 485
IOS see Input/Output Supervisor (IOS) responsibilities 398, 487
ioscan 322 master volume 183
ISMF 437 MAX_CHANNEL_IO 446
MAX_DEVICE_IO 446
max_initialization_rate 159–161
J max_resync_rate 159
JCL 444 max_syncjob_rate 159
JFS2 91–92 MAXIO 446
MAXTRK 446
K MDG 179
Keep Source Updated 136, 147, 158 MDisk 179
key-sequenced data set (KSDS) 450 modes 194
KSDS see key-sequenced data set (KSDS) member name 441
metainit 290
metastat 299–300, 303
L metattach 290, 300
label 95, 310 Metro Mirror 43, 176, 183, 188
last extent 195 link 44
lavailpprcport 63 monitor 254
LCU see logical control unit (LCU) performance 44
LDM see Windows Logical Disk Manager (LDM) Microsoft Cluster Server (MSCS) 148
license key 404, 418, 488 migrate a VDisk 185, 196
link 44 migrate between MDGs 185, 196
Linux x86 140 migrate data 194
listing utility 439 migratepv 329, 331, 334
Local Page 423 migratepv -p 336
logical control unit (LCU) 422 migration
Logical Disk Manager (LDM) 266 algorithm 185
Logical Volume Management 2 (LVM2) 376 group 397, 450, 485
Logical Volume Manager (LVM) 266 pair 397, 485
logredo 92 progress 198
LPAR 83, 396, 408, 420, 485 session 397, 485
lsfbvol 83 speed 137, 158
lslv 344 tips 195, 200
lspprc 72, 87–88 Mirror phase 435
lspprcpath 85 Mirror Synchronization phase 446
lsrcrelationship 256 mirroring 265
lssi 83 process 278
lsvg 333, 335, 344, 360 mkfbvol 82
lsvpcfg 334, 347, 381 mkflash 528
LUN ID 135, 145–146, 150, 334 mklv 344
LUN mapping 145 mklvcopy 329, 334, 339, 345
LUN numbering 140, 173 mkpprc 73
LUN0 140, 144, 171, 173 mkpprcpath 528
lvextend 328 mkrcconsistgrp 252
LVM see Logical Volume Manager (LVM) MPIO 267
LVM2 see Logical Volume Management 2 (LVM2) MSCS see Microsoft Cluster Server (MSCS)
lvreduce 328 Multi-Image Manager 457
Multipath Subsystem Device Driver (SDD) 182, 333, 377,
380
M multipathing 173
managed disk 179
MVS Write-to-Operator/ Write-to-Operator with Reply
managed disk group 179
(WTO/WTOR) 400
managed mode MDisk 194
management console 177
Master N
Console 178 naming convention 425
responsibilities 487 naming standard 436
SVC cluster 246 no source updating 135–136
Index 535
NOACKnowledge 400 RedHat 140
node 178 redirect phase 397
non-disruptive 266 reducevg 347
Non-Extended Mirrored phase 441 Refresh phase 401, 415
nonvolatile storage 43, 45 REFVTOC 443
non-VSAM volume record (NVR) 451 remote mirror pair 146
NVR see non-VSAM volume record (NVR) removing mirror 282
reserve handling 456
RESERVE macro 419
O resize 158, 164
ODM 91, 338, 348, 373 restore 10
offline data migration 185 restriction list 439
online data migration 188 resume phase 397
out of sync 46 reverse pacing 421, 428
bitmap 46 REXX 397, 486
ring complex 454
P RMC
pacing 421, 428 license code 51
pairing 425 pairs 70
parallelism 190 paths 54, 84
partial last extent 195 rmdev 338, 348
partition 295, 310, 381 RMFMON 422
PENDING.XD 75 rmkrcrelationship 253
per cluster 199 rmlv 345
per managed disk 199 rmpprc 74, 88
performance 44 rsExecuteTask 529
persistent allocation 444 RTM Resource Manager 432
physical extent 376
PID 291 S
planning 267 SAF see System Authorization Facility (SAF)
Point-in-Time Copy 522 SAN boot 134
point-in-time copy 145 SAN Volume Controller (SVC) 176
portperfshow 161 cluster 178
Post-Completion phase 435 Copy Services 183
PP 342 licence 176
PPRCOPY QUERY 76 Metro Mirror 183, 187–188
pre-migration 436 migration 180
preparation 267 migration algorithm 185
primary couple dataset 424 script 522–523
program temporary fix (PTF) 418 SCSI initiator 139
progress monitor 414 SDD command-line interface (SDDDSM) 277
PTF see program temporary fix (PTF) SDD see Multipath Subsystem Device Driver (SDD)
pvcreate 324, 382 SDDDSM see SDD command-line interface (SDDDSM)
PVID 90 SDSF see System Display and Search Facility (SDSF)
pvmove 376, 383 security 404, 436
profile 439
Q serialization 457
QLogic 380 SFM policy see sysplex failure management (SFM) policy
Quiesce phase 397, 401, 421, 485, 487 SFM see sysplex failure management (SFM)
showvolgrp 273, 334, 342
simplex 71
R slice 304
RACF 448 SMP/E 402
rank contention 422 SMS see Data Facility Storage Management Subsystem
raw device 9 (DFSMS)
READ authority 399 Solaris Volume Manager (SVM) 288
record type 451 mirroring process 289
recreatevg 90 source LUN 146, 158, 166
Redbooks website 531 Source Target System 146, 157
Contact us xviii source updating 135–136, 151, 169

option 135 47
source volume 396, 485 total cost of ownership (TCO) 434
special migration 195 TPC see Tivoli Storage Productivity Center (TPC) for
split mirroring 265 Replication
SSMzOS 452 track geometry 396, 485
stack 186 transfer rate 159
stale partition 346 transitions 194
star complex 456 Transparent Data Migration Facility (TDMF) for z/OS
started task 396 395–396, 483–484
startrcconsistgr 255 agent system 396, 485
startrcrelationship 256 responsibilities 398, 487
stmsboot 290, 292 architecture 397, 486
Storage Complex 54 best practices 418, 488
storage group 438 hardware 401, 487
storage subsystem 176 initialization 399
storage system 133–134 license key 404
storage virtualization 376 master system 396, 485
submirror 288, 297, 299 responsibilities 398, 487
SuSE 140 migration group 397, 485
suspend pairs 72 migration pair 397, 485
suspended 71 migration session 397, 485
SVC see SAN Volume Controller (SVC) process flow 398
svcinfo lsfreeextents 198 source volume 396, 485
svcinfo lsmdiskextent 198 swap migration 397, 485
svcinfo lsmigrate 198 system initialization 399
svcinfo lsVDiskextent 198 target volume 396, 485
svctask mkpartnership 249 TSO Monitor 431
SVM see Solaris Volume Manager (SVM) TSO 47, 103
swap migration 397, 405, 485
synchronization 151, 159, 397, 485
Synchronization phase 397, 401, 416, 435, 485 U
synchronized 135, 154, 159 umount command 92
synchronous reads 186 unfreezepprc 60
synchronous writes 186 Unidentified Systems 420
syncvg 329, 339, 345 uninterruptible power supply (UPS) 178
syslog 417 unmanaged MDisk 194
SYSOPTN 404, 418, 420 UPDATE authority 399
sysplex 455–456 UPS see uninterruptible power supply (UPS)
couple data set 456
sysplex failure management (SFM) 456 V
policy 456 VCMDB see Volume Configuration Management Data-
System Authorization Facility (SAF) 399, 405 base LUNs (VCMDB)
System Display and Search Facility (SDSF) 432 VDisks 176, 179
System Page Pool 423 image mode migration concept 194
System z 99 migration 181
interface 49 vendor identification (VID) 291
Veritas Storage Foundation 304
T Veritas Volume Manager (VxVM) 92, 266, 304–305
target 135, 137, 144 vgdisplay 324
target volume 396, 485 vgextend 325
target_list 152, 159 vgmodify 326
TCO see total cost of ownership (TCO) vgreduce 328, 386
TDMEXEC 410 vgscan 382
TDMF see Transparent Data Migration Facility (TDMF) for VID see vendor identification (VID)
z/OS virtual disks 179
Termination phase 401, 421 Virtual Storage Access Method (VSAM) 440
Test Data Migration 148–149, 154, 158 volser 439
thick to thin migration 162 Volume Configuration Management Database LUNs (VC-
thin provisioning 162 MDB) 294, 307
Tivoli Storage Productivity Center (TPC) for Replication volume table of contents (VTOC) 450, 458
Index 537
volumes
acknowledgement 400
confirmation 400
I/O redirect phase 401
initialization 401
pair assignment 66
selection 401
vpath 342, 347, 381–382
VPSMS 444
VSAM see Virtual Storage Access Method (VSAM)
VSAM volume data set (VVDS) 450, 458
VSAM volume record (VVR) 451
VTOC see volume table of contents (VTOC)
VTOCIX 440
VVDS see VSAM volume data set (VVDS)
VVR see VSAM volume record (VVR)
vxdg 98, 306
vxdisk list 310
vxdiskadm 306, 312, 314
vxplex 306, 318
vxprint 308, 319
vxtask 317
VxVM see Veritas Volume Manager (VxVM)
W
Windows Logical Disk Manager (LDM) 268
worldwide port name (WWPN) 59, 140, 142, 144, 153,
166
write access 258
WTO/WTOR see MVS Write-to-Operator/ Write-to-Oper-
ator with Reply (WTO/WTOR)
WWPN see worldwide port name (WWPN)
X
XCF see cross-system coupling facility (XCF)
XCLI see XIV command-line interface (XCLI)
XIV command-line interface (XCLI) 152
XIV Storage System 133–134
Z
z/OS Dataset Mobility Facility (zDMF) 433
parameters 446
performance 445
storage requirements 447
zDMF see z/OS Dataset Mobility Facility (zDMF)
zone 288, 338, 379
zoning 139–140, 149, 165

Data Migration to IBM Disk Storage
Systems
Systems
(1.0” spine)
0.875”<->1.498”
460 <-> 788 pages
Systems
Systems
Back cover ®
Data Migration to IBM

Disk Storage Systems
®
Highlights tool and Data migration has become a mandatory and regular activity for
most data centers. Companies need to migrate data not only when INTERNATIONAL
techniques for Open
technology needs to be replaced, but also for consolidation, load TECHNICAL
Systems and z/OS
balancing, and disaster recovery. SUPPORT
This IBM Redbooks publication addresses the aspects of data ORGANIZATION
Addresses appliance,
host, and migration efforts while focusing on the IBM System Storage as the
target system. Data migration is a critical and complex operation,
storage-based
and this book provides the phases and steps to ensure a smooth
migrations migration. Topics range from planning and preparation to execution
BUILDING TECHNICAL
and validation.
Includes z/OS TDMF, INFORMATION BASED ON
The book also reviews products and describes available IBM data PRACTICAL EXPERIENCE
TDMF TCP/IP, and migration services offerings. It explains, from a generic standpoint,
zDMF the appliance-based, storage-based, and host-based techniques IBM Redbooks are developed
that can be used to accomplish the migration. Each method is by the IBM International
explained including the use of the various products and techniques Technical Support
with different migration scenarios and various operating system Organization. Experts from
IBM, Customers and Partners
platforms. from around the world create
This document targets storage administrators, storage network timely technical information
administrators, system designers, architects, and IT professionals based on realistic scenarios.
who design, administer or plan data migrations in large data Specific recommendations
Centers. The aim is to ensure that you are aware of the current
are provided to help you
implement IT solutions more
thinking, methods, tools, and products that IBM can make available effectively in your
to you. These items are provided to ensure a data migration environment.
process that is as efficient and problem-free as possible.
The material presented in this book was developed with versions of
the referenced products as of June 2011.
For more information:
ibm.com/redbooks
SG24-7432-01 ISBN 0738436283

Data Migration To IBM Storage Systems PDF

Uploaded by

Copyright:

Available Formats

Data Migration To IBM Storage Systems PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Data Migration To IBM Storage Systems PDF

Uploaded by

Copyright:

Available Formats

Front cover

Data Migration to IBM

Addresses appliance, host, and

Includes z/OS TDMF, TDMF

Data Migration to IBM Disk Storage Systems

Second Edition (February 2012)

Summary of changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii

Part 1. Data migration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

Chapter 1. Introducing disk data migration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

Chapter 2. Migration techniques and processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

Chapter 3. IBM service offerings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

© Copyright IBM Corp. 2011, 2012. All rights reserved. iii

Part 2. Migration scenarios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

Chapter 4. Data migration using IBM Remote Mirror and Copy . . . . . . . . . . . . . . . . . . 41

Chapter 5. DSCLIbroker . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105

iv Data Migration to IBM Disk Storage Systems

Chapter 6. IBM XIV data migration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133

Chapter 7. SAN Volume Controller-based migration . . . . . . . . . . . . . . . . . . . . . . . . . . 175

Chapter 8. Using mirroring techniques. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265

vi Data Migration to IBM Disk Storage Systems

Chapter 9. Using TDMF for z/OS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 395

Chapter 10. z/OS Dataset Mobility Facility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 433

Chapter 11. Using TDMF TCP/IP for z/OS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 483

viii Data Migration to IBM Disk Storage Systems

Appendix A. Network block devices for Linux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 515

Appendix B. CLI Conversion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 521

Related publications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 531

© Copyright IBM Corp. 2011, 2012. All rights reserved. xi

The following terms are trademarks of other companies:

xii Data Migration to IBM Disk Storage Systems

February 2012, Second Edition

© Copyright IBM Corp. 2011, 2012. All rights reserved. xiii

This document targets storage administrators, storage network administrators, system

The team who wrote this book

© Copyright IBM Corp. 2011, 2012. All rights reserved. xv

xvi Data Migration to IBM Disk Storage Systems

Thanks to the following people for their contributions to this project:

Bernd Mueller, Michael Murtagh, Ralf Wohlfarth

Lori Bideaux and Bert Dufrasne

Now you can become a published author, too!

Stay connected to IBM Redbooks

xviii Data Migration to IBM Disk Storage Systems

Part 1 Data migration

This part contains the following chapters:

© Copyright IBM Corp. 2011, 2012. All rights reserved. 1

Chapter 1. Introducing disk data migration

© Copyright IBM Corp. 2011, 2012. All rights reserved. 3

You might migrate data for the following reasons:

1.2 Aspects of data migration

4 Data Migration to IBM Disk Storage Systems

Chapter 1. Introducing disk data migration 5

Chapter 2. Migration techniques and

This chapter includes the following sections:

© Copyright IBM Corp. 2011, 2012. All rights reserved. 7

Various software products and tools can be used to migrate data:

Determine whether you can migrate data online or offline:

2.1.1 Host-Based migration

Volume management software

8 Data Migration to IBM Disk Storage Systems