Perf Reporting

Download as pdf or txt
Download as pdf or txt
You are on page 1of 32

Understanding and Creating

Effective Performance Test Reports

First Presented for:


Florida Institute of Technology
Computer Science Seminar

Scott Barber
Chief Technology Officer
PerfTestPlus, Inc.

www.PerfTestPlus.com Effective Performance Test Reports Page 1


© 2006 PerfTestPlus, Inc. All rights reserved.
Agenda
Introduction
What Stakeholders Want
What Stakeholders (Usually) Get
How to “give ‘em what they want” (or at least get close)
Valuable Charts, Tables, Verbiage and Their Uses
Stakeholder Reaction to Charts, Tables and Verbiage
Comprehensive Final Report Template
Want More Information?
Summary/Questions

www.PerfTestPlus.com Effective Performance Test Reports Page 2


© 2006 PerfTestPlus, Inc. All rights reserved.
Introduction
Most people will never read performance test results docs.
Most people don’t really understand the underlying
components to performance.
It is our job to make it easy for them to understand, and
understand quickly.
Being skilled at graphical presentation of technical
information is critical for us to help others understand the
message we are delivering.
Improper or confusing charts and tables can lead to wrong
decisions that cause lost $ and ruined reputations.

www.PerfTestPlus.com Effective Performance Test Reports Page 3


© 2006 PerfTestPlus, Inc. All rights reserved.
What Stakeholders Want
Answers… NOW! (They might not even know the question)

To understand information intuitively.


Simple explanations of highly technical information.
To be able to make decisions quickly and have the
information to support those decisions.
“Trigger phrases” to use with other stakeholders.
Concise summaries and conclusions.
Recommendations and options.

www.PerfTestPlus.com Effective Performance Test Reports Page 4


© 2006 PerfTestPlus, Inc. All rights reserved.
What Stakeholders (Usually)
Get (verbal) technical descriptions.
Complex
Complex technical graphics.
Raw data.
i.e. they get confused!

www.PerfTestPlus.com Effective Performance Test Reports Page 5


© 2006 PerfTestPlus, Inc. All rights reserved.
“Give ‘em what they want” (or at least get
close)
Concise verbal descriptions.
Well formed, informative charts (pretty pictures).
Focus on requirements and business issues.
Don’t be afraid to make recommendations or draw
conclusions!
Have supporting data available.

www.PerfTestPlus.com Effective Performance Test Reports Page 6


© 2006 PerfTestPlus, Inc. All rights reserved.
Charts, Tables and Verbiage
Performance Report Output Chart
Response vs. Time Scatter Chart
Component (Resource Usage) Performance Chart(s)
Consolidated Scatter Chart
Response Time Summary Comparison Chart
Response Time by Test Execution Chart
Response Time Degradation Curve

www.PerfTestPlus.com Effective Performance Test Reports Page 7


© 2006 PerfTestPlus, Inc. All rights reserved.
Performance Report Output Chart

“This chart shows that the home page meets the 6 second
response time requirement 90% of the time under the
tested user load. The requirements demand 95%
compliance. Page1 was the next slowest page, but it
achieved the required response time.”

www.PerfTestPlus.com Effective Performance Test Reports Page 8


© 2006 PerfTestPlus, Inc. All rights reserved.
Performance Report Output Chart
Good for:
• Showing response times for high interest pages during a single
test.
• Highlighting select instances of poor (or good) performance.
Not good for:
• Results from multiple tests.
• Component/resource based info.
• Results about many pages at once.
Notes:
• This chart and table can be viewed directly in TestManager or
recreated in Excel.

www.PerfTestPlus.com Effective Performance Test Reports Page 9


© 2006 PerfTestPlus, Inc. All rights reserved.
Response vs. Time Scatter Chart
“This chart shows that the home page (blue squares) is the
slowest page with most response times hovering around
the the 6 sec requirement. Page1 (purple X) was the next
slowest page, but it achieved the required response time.”

www.PerfTestPlus.com Effective Performance Test Reports Page 10


© 2006 PerfTestPlus, Inc. All rights reserved.
Response vs. Time Scatter Chart
Good for:
• Identifying patterns in response times over a whole run.
• Graphically displaying response times vs. goals.
• Highlighting select instances of poor performance.
• Can be overlayed with component/resource data.
Not good for:
• Results from multiple tests.
Notes:
• This chart and table can be viewed directly in TestManager or
recreated in Excel.
• Many component/resource metrics can be collected and
overlayed directly in TestManager.

www.PerfTestPlus.com Effective Performance Test Reports Page 11


© 2006 PerfTestPlus, Inc. All rights reserved.
Component Performance Chart

“The red line shows that the CPU of the web server
bounced between 0 and 100% utilization during a test
run that was not designed to be stressful. This is
abnormal and needs to be researched further.”
www.PerfTestPlus.com Effective Performance Test Reports Page 12
© 2006 PerfTestPlus, Inc. All rights reserved.
Component Performance Chart
Good for:
• Identifying various resource usage.
• Determining issues such as memory leaks.
Not good for:
• Response time reporting.
• Results from multiple tests.
Notes:
• Charts like this one can be created during your script by putting
agents on the servers to be monitored.
• Charts like this one can also be created using tools like Perfmon,
Perfmeter and Top.

www.PerfTestPlus.com Effective Performance Test Reports Page 13


© 2006 PerfTestPlus, Inc. All rights reserved.
Consolidated Scatter Chart
“This chart shows both response times and resource
utilization together. Close examination shows that
Application Server CPU Usage and Queue length
coincide with degraded Response Time.”

www.PerfTestPlus.com Effective Performance Test Reports Page 14


© 2006 PerfTestPlus, Inc. All rights reserved.
Response vs. Time Scatter Chart
Good for:
• Identifying correlations between response times and resource
usage during over time.
• Graphically displaying response times vs. resources.
• Highlighting potential causes of poor performance.
• Technical stakeholders.
Not good for:
• Results from multiple tests.
• “Low tech” stakeholders.
Notes:
• This chart and was created in Excel, but in many cases can be
done directly in TestManager.

www.PerfTestPlus.com Effective Performance Test Reports Page 15


© 2006 PerfTestPlus, Inc. All rights reserved.
Response Time Summary Comparison
“This chart shows that the majority of pages met their
required response times at up to 150 hourly users
accessed via LAN, but that most did not via slower
connections.”

www.PerfTestPlus.com Effective Performance Test Reports Page 16


© 2006 PerfTestPlus, Inc. All rights reserved.
Response Time Summary Comparison
Good for:
• Summarizing requirements met across various and diverse tests.
• Comparing results across multiple tests.
• Highlighting tests with particularly bad (or good) performance.
Not good for:
• Depicting individual response times or component/resource data.
Notes:
• This chart and table must be created by hand in Excel using test
execution data and stated requirements.

www.PerfTestPlus.com Effective Performance Test Reports Page 17


© 2006 PerfTestPlus, Inc. All rights reserved.
Response Time by Test Execution

“This chart shows that response times for both Home Page
and Page1 degrades dramatically and unacceptably at
200 hourly users and via modem connection speeds.”
www.PerfTestPlus.com Effective Performance Test Reports Page 18
© 2006 PerfTestPlus, Inc. All rights reserved.
Response Time by Test Execution
Good for:
• Comparing specific page response times across multiple tests.
• Graphically displaying scenarios performance side-by-side.
• Highlighting test configurations with universally poor performance.
Not good for:
• Displaying component/resource data.
Notes:
• This chart and table must be created in Excel from data exported
from TestManager.

www.PerfTestPlus.com Effective Performance Test Reports Page 19


© 2006 PerfTestPlus, Inc. All rights reserved.
Response Time Degradation Curve
“This chart shows that performance starts to degrade at 150
hourly users and becomes unacceptable between 175
and 200 hourly users.”

www.PerfTestPlus.com Effective Performance Test Reports Page 20


© 2006 PerfTestPlus, Inc. All rights reserved.
Response Time Degradation Curve
Good for:
• Comparing specific page response times across multiple tests.
• Graphically displaying where performance gets bad.
• Highlighting load where performance becomes unacceptable.
Not good for:
• Displaying component/resource data.
Notes:
• Generally considered the most powerful graph at a performance
tester/engineer’s disposal.
• This chart and table must be created in Excel from data exported
from TestManager.

www.PerfTestPlus.com Effective Performance Test Reports Page 21


© 2006 PerfTestPlus, Inc. All rights reserved.
Response Time Degradation Curve
Single User Region:
• Typically slightly slower than best case.
• Think of it like “warming up your car on a cold day.”

www.PerfTestPlus.com Effective Performance Test Reports Page 22


© 2006 PerfTestPlus, Inc. All rights reserved.
Response Time Degradation Curve
Performance Plateau:
• The best performance you can expect without further tuning.
• Good candidates for baselines and/or benchmarks.

www.PerfTestPlus.com Effective Performance Test Reports Page 23


© 2006 PerfTestPlus, Inc. All rights reserved.
Response Time Degradation Curve
Stress Region:
• Area where the application “degrades gracefully”.
• Max recommended user load is the beginning of the stress
region.

www.PerfTestPlus.com Effective Performance Test Reports Page 24


© 2006 PerfTestPlus, Inc. All rights reserved.
Response Time Degradation Curve
Knee in Performance:
• Point where performance degrades “un-gracefully”.
• There is always a knee, it’s our job to find it so we can plan
around it.

www.PerfTestPlus.com Effective Performance Test Reports Page 25


© 2006 PerfTestPlus, Inc. All rights reserved.
Response Time Degradation Curve
Min/Max/Expected Version:
• Comparing performance based on various workload distributions.
• Yields a “confidence interval for predicted performance.”

www.PerfTestPlus.com Effective Performance Test Reports Page 26


© 2006 PerfTestPlus, Inc. All rights reserved.
www.PerfTestPlus.com Effective Performance Test Reports Page 27
© 2006 PerfTestPlus, Inc. All rights reserved.
Stakeholder Reactions
“These are great, but where’s the supporting data?”
• All the data lives in TestManager and/or Excel.
• Include the spreadsheets as appendices with final report.
“Very pretty, but what do they mean?”
• That’s why we always have a paragraph that tells what the
chart/table means, not what it says.
“Terrific! This is exactly what I wanted! Don’t worry about the
final report — these will do nicely.”
• As tempting as it may be – don’t do it. Write the report anyway…
trust me.

www.PerfTestPlus.com Effective Performance Test Reports Page 28


© 2006 PerfTestPlus, Inc. All rights reserved.
Comprehensive Final Report Template
Executive Summary!!!
Re-cap Acceptance Criteria (Requirements)
Re-Cap Workloads Actually Used
Summarize Tests Individually
Summarize Tuning Activity (System Modifications)
Summarize Results Collectively
Document Conclusions
Document Recommendations

www.PerfTestPlus.com Effective Performance Test Reports Page 29


© 2006 PerfTestPlus, Inc. All rights reserved.
Want More Information?
Information adapted from User Experience, not Metrics: Parts 6,
8, 9 and10 and Beyond Performance Testing: Part 2 located at
http://www.rational.net (RDN) and
http://www.PerfTestPlus.com (My site)
Good sources for additional information about Performance
Testing:
• http://www.PerfTestPlus.com (Methodology, Templates, Articles,
Presentations)
• http://www.loadtester.com (Good articles and links)
• http://www.keynote.com/resources/resource_library.html (Good articles
and statistics)
Graphical Presentation of Information – Edward Tufte, PhD.
http://www.edwardtufte.com (Books and seminars)

www.PerfTestPlus.com Effective Performance Test Reports Page 30


© 2006 PerfTestPlus, Inc. All rights reserved.
www.PerfTestPlus.com Effective Performance Test Reports Page 31
© 2006 PerfTestPlus, Inc. All rights reserved.
Contact Info

Scott Barber
Chief Technology Officer
PerfTestPlus, Inc

E-mail: Web Site:


[email protected] www.PerfTestPlus.com

www.PerfTestPlus.com Effective Performance Test Reports Page 32


© 2006 PerfTestPlus, Inc. All rights reserved.

You might also like