DATA CENTRE NETWORKED APPLICATIONS BLUEPRINT

A KEY FOUNDATION OF CISCO SERVICE-ORIENTED NETWORK ARCHITECTURE

SECONDARY INTERNET EDGE AND EXTRANET

LARGE BRANCH OFFICE INTERNET EDGE
EXTRANET Perform Distributed Denial of Service (DDoS) Partners
Use dual Integrated Services Routers for a large attack mitigation at the Enterprise Edge. Place the
branch office. Each router is connected to different Use a collapsed Internet Edge and
Guard and Anomaly Detector to detect and extranet design for a highly centralized
WAN links for higher redundancy. Use dual stack mitigate high-volume attack traffic by diverting
IPv6-IPv4 services on Layer 3 devices. Use the and integrated edge network. Edge
Use a dedicated extranet as a highly through anti-spoofing and attack specific dynamic services are provided by embedding
integrated IOS firewall and intrusion prevention for scalable and secure termination point filter counter-measures. Use the Adaptive Security
edge security and the integrated Call Manager for intelligence from service modules such as
for IPsec and SSL VPNs to support Appliance to concurrently perform firewall, VPN, firewall, content switching and SSL (ACE)
remote voice capabilities. Consider IPv6 firewall business partner connectivity. Apply and instrusion protection functions at the edge of
policies, filtering and DHCP prefix delegation when and VPN modules, and appliances such
IPv6 traffic is expected to/from branch offices. Use
the intranet server farm design best
practices to the specific partner
the enterprise network. Use dual-stack IPv6-IPv4
on the edge routers and consider IPv6 firewall
Service Service as Guard XT and the Anomaly Detector for
DDoS protection. Additional edge
integrated Wide Area Engines for file caching,
local video broadcast, and static content serving.
facing application environments, but
considering their specific security and
and filtering capabilities. Provider 1 Provider 2 functionality includes site selector and
content caching, as well as event
scalability requirements. Consider the correlation engines and traffic monitoring
use of the Application Control Engine Place the Global Site Selector in the DMZ to provided by the integrated service devices.
Large branches have similar LAN/SAN designs to

INTERNET
(ACE) to provide high performance prevent DNS traffic from penetrating the edge Consider the use of dual-stack IPv4-IPv6
small data centers and small campuses for server and high scalability load balancing, security boundaries. Consider the design of the services in Layer 3 devices and the need
and storage, and client connectivity respectively. SSL and application security Internet-facing server farm following the same to support IPv6 firewall policies and IPv6
Use Layer 3 switches to house branch-wide LAN
services and to provide connectivity to all required
capabilities per partner profile by using
virtual instances of these functions.
best practices used in intranet server farms, with
specific scalability and security requirements
VPN filtering capabilities.
access layer switches. Consider a number of
driven by the size of the target user population.
VLANs based on the branch functions such as
Use the AONs application gateway for XML
server farm, point of sale, voice, video, data,
filtering such as schema and digital signature
wireless and management in the LAN design.
validation, and provide transaction integrity and
security for legacy application message formats.

Consider the storage network design and available

storage connectivity options: FC, iSCSI and NAS.
Plan the data replication process from the branch
to headquarters based on latency and transaction
rates requirements. Consider QoS classification to SECONDARY CAMPUS NETWORK
ensure the different types of traffic match the loss
and latency requirements of the applications. CAMPUS CORE

CAMPUS NETWORK Integrate wireless controllers at

Building Y WIDE AREA NETWORK the distribution layer and wireless
access points at the access layer.
When Layer 2 is used in the
Campus access layer, select
CAMPUS CORE Use Etherchannel between the
distribution switches to provide
a primary distribution switch redundancy and scalability.
to be the primary default
gateway and STP root.
The Campus core provides connectivity between PSTN Dual-home access switches to the
distribution layer to increase
REMOTE OFFICES Set the redundant distribution
the major areas of an Enterprise network including
the data center, extranet, Internet edge, Campus, redundancy by providing alternate
switch as the backup Wide Area Network (WAN), and Metropolitan Area paths. Consider the use of
default gateway and Network (MAN). Use a fully-meshed Campus core dual-stack IPv4-IPv6 services at
HOME OFFICE SMALL OFFICE
secondary root. Use to provide high-speed redundant Layer 3 the access, distribution and core
HSRP and PVRST+ connectivity between the different network areas. Layer 3 and/or Layer 2 devices.
as the primary default Use dual-stack IPv6-IPv4 in all Layer 3 devices
Consider an integrated services design gateway and STP protocols. and desktop services.
Consider the home office as an
extension of the enterprise network. for a full service branch environment.
Basic services include access to Services include voice, video, security PRIMARY SITE SECONDARY SITE Building Z
applications and data, voice and and wireless. Voice services include IP
video. Use VPN to ensure security phones, local call processing, local voice
for teleworker environments, thus mail, and VoIP gateways to the PSTN.
relying on the corporate security Security services include integrated
policies. Also consider the use of firewall, intrusion protection, IPsec and
wireless access as an extension of admission control. Connect small office Use 10GbE throughout the
SECONDARY DATA CENTER
Use the WAN as the primary path for user traffic
the enterprise network; secure and networks to headquarters through VPN, infrastructure (between destined for the intranet server farm. Through the use
reliable. and ensure QoS classification and distribution switches and of DNS and RHI control the granularity of applications
Enable QoS to police and enforce enforcement provides adequate service between access and being independently advertised, and state of distributed
service levels for voice, data and levels to the different traffic types. distribution) when high application environments. Ensure the proper QoS
video traffic. Consider a dual-stack Configure multicast for applications that throughput is required. classification is used for voice, data and video traffic. Use the secondary data center as a
IPv6-IPv4 router to support IPv6 require concurrent recipients of the same Use Layer 3 access Use dual-stack IPv6-IPv4 in all Layer 3 devices. backup location that houses critical
remote devices. Security policies for traffic. Consider a dual-stack IPv4-IPv6 switches when shared standby transactional (near zero
IPv6 traffic should include IPv6 router to support IPv6 traffic. Ensure VLANs are not needed in RPO and RTO) and redundant active
filtering and capable firewalls. IPv6 firewall rules and filtering more than one access non-transactional applications (RPO
capabilities are enabled on the router. switch at a time, and very and RTO in the 12-24 hours range).
low convergence is required.

Building X The secondary data center design is

a smaller replica of the primary that
houses backup critical application
environments. These support business
functions that must be resumed to
achieve regular operating conditions.

LARGE-SCALE PRIMARY DATA CENTER BLADE SERVER COMPLEX

MAN INTERCONNECT Group servers providing like-functions
in the same VLANs to apply consistent
and manageable set of security, SSL,
METRO ETHERNET
COLLAPSED MULTITIER DESIGN DATA CENTER CORE Place all network-based service devices (modules or
load balancing, and monitoring policies.
Dual-home critical servers to different
appliances) at the aggregation layer to centralise the
access switches, and stagger primary
configuration and management tasks and to leverage
Use a data center core layer to support multiple aggregation physical connections between
Use ACE as a content switch to service intelligence applied to the entire server farm.
modules. Use multiple aggregation modules when the number Use a high-speed (10GbE) metro optical available access switches.
scale application services including
of servers per module exceeds the capacity of the module. network for packet-based and transparent
SSL off-loading on server farms. Consider blade server direct attachment and network
Connect the data center core switches to the campus core LAN services between distributed Campus
Use virtual firewalls to isolate fabric options; Pass-through modules or integrated Use PortChannels and trunks to
switches to reach the rest of the Enteprise network. Consider and Data Centre environments.
application environments. Use AON switches, and Fibre Channel, Ethernet and Infiniband. Blade Blade aggregate multiple physical inter-switch
the use of 10GbE links between core and aggregation switches. Servers Servers
to optimise inter-application security links (ISL) into a logical link. Use
Use dual-stack IPv6-IPv4 in all Layer 3 devices in the data
and communications services, and In an integrated Ethernet switch fabric, set up half the VSANs to segregate multiple distinct
center, and identify the server farm requirements to ensure
to provide visibility into real-time blades active on switch1 and half active on switch2. A SANs in a physical fabric to consolidate
IPv6 traffic conforms to firewall and filtering policies. Infiniband B
transactions. Use MARS to detect Dual-home each Ethernet switch to Layer 3 switches isolated SANs and SAN fabrics. Use
security anomalies by correlating through GbE-channels. Use RPVST+ for fast STP Network core-edge topologies to connect
data from different traffic sources. convergence. Use link-state tracking to detect uplink multiple workgroup fabric switches
failure and allow the blades standby NIC to take over. when tolerable over-subcription is a
design objective.
Attach integrated Infiniband switches to Server Fabric
Consolidate application and Switches acting as gateways to the Ethernet network.
security services (service modules EXPANDED MULTI-TIER DESIGN Connect the gateway switches to the aggregation
switches to reach the IP network.
Use storage virtualisation to further
increase the effective storage utilization
or appliances) at the aggregation
layer switches. Ensure the access and centralise management of storage
A B arrays. Arrays form a single pool of
layer design (whether L2 or L3) Application, security and virtualisation When using pass-through modules dual-home
provides a predictable and servers to access/edge layer switches. Pass-through virtual storage which are presented
services provided by service modules
deterministic behavior and allows modules allow Fibre Channel environments to avoid as virtual disks to applications.
or appliances are best offered from the Use a SONET/SDH transport network for FCIP,
the server farm to scale up the aggregation layer. Services are made interoperability issues while allowing access to the in addition to voice, video, and additional IP
expected number of nodes. Use available to all servers, provisioning is advance SAN fabric features. traffic between distributed locations in a metro
VLANs in conjunction with centralized and the network topology or long-haul environments. Consider the use of
instances of application and is kept predictable and deterministic. RPR/802.17 technology to create a highly
security services applied to each
4992 Node Ethernet Cluster
available MAN core for distributed locations.
application environment
independently. Select a primary aggregation switch to HIGH DENSITY ETHERNET CLUSTER • Modular chassis per rack group
A B be the primary default gateway and • 8 aggregation - 16 access switches
STP root. Set the redundant High density Ethernet clusters consist of
• 16 10GbE downlinks per aggregation
aggregation switch as the backup multiple servers that operate concurrently
• 8 10GbE uplinks per access
Use VSANs to create separate default gateway and secondary root. to solve computational tasks. Some of
• Layer 3 in aggregation and access
SANs over a shared physical Use HSRP and RPVST+ as the these tasks require certain degree of
• 8-way equal cost multipath ECMP
infrastructure. Use two distinct SAN primary default gateway and STP processing parallelism while others require
• 312 GbE ports per access switch
fabrics to mainain a highly available protocols and ensure he active service raw CPU capacity. Common applications
• 3.9:1 Oversubscription
SAN environment. Use port channel devices are in the STP root switch. of large Ethernet clusters include large
• 80 Gigabit per access switch
to increase path redundancy and search engine sites and large web server
fast recovery from link failure. Use farms. The diagram shows a tiered design
The core layer is required when the
FSPF for equal cost load-balancing Deploy access layer switches in pairs using “top of rack” 1RU access switches
cluster needs to connect to an existing
through redundant paths. Use to enable server dual-homing and NIC for a total of 1536 servers.
IP network environment. The modular
storage virtualisation to pool distinct Teaming. Use trunks and channels
access layer switches provide access
physical storage arrays as one, between access and aggregation Topology Details:
Use a DWDM/SONET/SDH/Ethernet transport 4992 GbE attached servers functions to groups of racks at a time.
hiding physical details (arrays, switches. Carry VLANs that are needed
network to support high-speed, low-latency Design is aimed at reducing hop count
spindles, LUNs). throughout the server farm on every - 8 core switches connected to each
uses, such as synchronous data replication between any two nodes in the cluster.
trunk, to increase flexibility. Trimm aggregation module through a 10GbE
unneeded VLANs from every trunk. link per switch between distributed disk subsystems. The
- 4 aggregation modules each with 2 common transport network supports multiple
Layer 3 switches that provide 10GbE protocols such as FC, GbE, and ESCON
Use firewalls to control the traffic path connectivity to the access layer switches concurrently. 288 Node Infiniband Cluster
HIGH PERFORMANCE INFINIBAND CLUSTER between tiers of servers and to isolate - 8 access switches per aggregation
SERVER module, each switch connecting to 2 Use a non-blocking design for server
distinct application environments. Use
CLUSTER aggregation switches through 10GbE clusters dedicated to computational tasks.
Use an Infiniband fabric for applications that execute a high ACE as a content switch to monitor SERVER
VSAN links
1536 GbE In a non-blocking design for every HCA
rate of computational tasks and require low latency and high and control server and application servers CLUSTER
- Each access layer switch supports 48 connected to the edge/access layer, there
throughput. Select the proper oversubcription rate between health, and to distribute traffic load VSAN
10/100/1000 ports and 2 10GbE uplinks is an uplink to an aggregation layer switch.
edge and aggregation layers for intracluster purposes, or between clients and the server farm,
between aggregation, core and the Ethernet fabric. In a 2:1 and between server/application tiers.
Topology Details
blocking topology for every two HCAs connected to edge Edge
switches, there is an uplink to an aggregation layer switch. Core To achieve additional redundancy on an HA server cluster, distribute a portion of • 24 switches - 24 ports each
switches provide connectivity to the Ethernet fabric. Use Vframe Connect access switches used in the servers in the HA cluster to a data center. This distribution of HA clusters • 12 servers per switch
to manage I/O virtualisation capabilities of server fabric switches. application and back-end segments to HIGH AVAILABILITY SERVER CLUSTER across distributed data centers, referred to as geo-clusters or stretched clusters, • 12 uplinks to aggregation layer
each other across application tier often times requires Layer 2 adjacency between distributed nodes. Adjacency Aggregation
Topology details for 1024 servers include:
function boundary through means the same VLAN (IP subnet) and VSAN have to be extended over the • 12 switches - 24 ports each
Edge: EtherChannel® links. Use VLANs to High availability clusters consist of multiple
shared transport infrastructure, between the distributed data centers. The HA • 1 or more uplinks to each core switch
• 96 switches - 24 ports each - 12 servers per switch separate groups of servers by function servers supporting mission-critical applications
cluster spans multiple geographically distant data center hosting facilities. Core
• 12 Uplinks to aggregation layer or application service type. in business continuance or disaster recovery
scenarios. The applications include databases, • number of core switches based on
Aggregation: connectivity needs to IP network
• 6 switches - 96 ports each filers, mail servers or file servers. The nodes of
• 12 downlinks - one per edge switch a single application cluster use a clustering
Use VSANs to group isolated fabrics
Core: into a shared infastructure while A B mechanism that relies on unicast packets if PUBLIC VLANs
• Number of switches based on fabric connectivity needs there are two nodes or multicast if using more
keeping their dedicated fabric HIGH AVAILABILITY SERVER CLUSTER
than two nodes. The nodes backup
services, security, and stability The nodes in HA clusters are linked
each other and use heartbeats to determine
integral per group. Dual-home hosts CLUSTER VLANs to multiple networks using existing
node status. The network infrastructure
to each of the SAN fabrics using Fibre network infrastructure. Use the
suporting the HA cluster is shared by other
Channel Host Bus Adapters (HBAs). private network for heartbeats and
NETWORK OPERATIONS CENTER (NOC) server farms. Additional VLANs and VSANs
the public network for inter-cluster
NOC VLAN/VSAN are used to connect additional NICs and HBAs FABRIC C communication and client access.
required by the cluster to operate. The
Use a dual-fabric (fabrics A and B) Nodes in distributed data centers
application data must be available to all nodes
topology to achieve high resiliency in may need to be in the same subnet,
in the cluster. This requires the disk to be shared
SAN environments. A common FABRIC D requiring Layer 2 adjacency.
so it cannot be local to each node. The shared
management VSAN is recommended disk or disk array is also accessible through IP
to allow the fabric manager to manage (iSCSI or NAS), Fibre Channel (SAN) or shared
and monitor the entire network VLAN X VLAN Y VSAN P VSAN Q
SCSI. The transport technologies that can be
Use a NOC VLAN to house critical management tools and to isolate management traffic from client/server traffic. Use NTP, environment. used to connect the LAN and the SAN of the PUBLIC PRIVATE
SSH-2, SNMPv3, CDP and Radius/TACACS+ as part of the management infrastructure. Use CiscoWorks LMS to manage data centers can be Dark Fiber, DWDM, CWDM, The transport network supports multiple communication streams between nodes NETWORK
NETWORK
the network infrastructure and monitor IPv4-IPv6 traffic, and the Cisco Security Manager to control, configure and deploy SONET, Metro Ethernet, EoMPLS, L2TPv3 as in the stretched clusters. Use multiple VLANs to separate intracluster from client
firewall, VPN and IPS security policies. Use on the Performance Visibility Manager to measure end-to-end application some of the options shown above. traffic. Use multiple SAN fabrics to provide path redundancy for the extended
performance. Use the Monitoring, Analysis, and Response System to correlate traffic for anomaly detection purposes. Use SAN. Use multipathing on the hosts and IVR between the SAN fabric Directors to
the Network Planning Solution to build network topology models, for failure scenario analysis and other what-if scenarios take advantage of the redundant fabrics. Use write acceleration to improve the
based on device configuration, routing tables, NAM and NetFlow data. Use the MDS Fabric Manager to manage the storage performance rate of the data replication process. Consider the use of encryption
network. Use NetFlow and the Network Analysis Module for capacity planning and traffic profiling. to secure data transfers and compression to increase the data transfer rates.

Designed By: Data Center Fundamentals: Service

Modules End-user
10 GbE
Workstation GbE

Mauricio Arregoces www.ciscopress.com/datacenterfundamentals Area Wireless

Connection

[email protected]
Service Service Cisco Cisco 3000
Questions: Design Best Practices: Appliances Devices
Placement
Application
Control
Series
Area Multifabric

[email protected]
www.cisco.com/go/datacenter Location Engine
Server Switch

Part #: 910300406R01