Load Testing is Easy, Good Load Testing is Not. Preparation is the Difference.

1
Load Testing is Easy, Good Load Testing is Not.
Preparation is the Difference.

• Chris Lynn - Safeway, Inc.
• Dennis Sherman - Safeway, Inc.

2
Disclaimers

• There will be no Greek letters in this
  presentation.
• This is an INTRODUCTORY starting point, not an
  all inclusive list of everything.
• If you are an advanced load tester then this
  might be therapy for you to see the same problems
  in other organizations.

3
Who is the audience?

• How many of you have participated in load testing
  for 1, 3, 5, 10 years?
• How many of you have used/use currently
  LoadRunner, QALoad, eTester, SLAMD, Load Copient,
  other?
• What percentage of projects you have load tested
  passed all performance and throughput
  requirements without any rework of the
  application? (1, 5, 10, 20)
• How Many are Load Testers, Capacity Planners,
  Other?

4
How do you feel?

• Often surprised (the bad kind) by my production
  performance when a new application is released?
• I am missing something in my testing?
• Production performance doesnt match the
  performance in testing?
• Do you feel lucky, punk?

5
What is the Purpose of Testing?What do you want
to learn?

• Performance Evaluation
• Performance Tuning
• Stress
• Stability
• Availability
• Avoid re-work and wasted work

6
Who Is Involved? Roles/Responsibilities

• Test Manager
• Load Tester
• Application/Business Representatives
• Capacity Planner
• Server Engineer
• Network Engineer

7
Know The Business Requirements (not Business
Hopes)

• SLAs
• Transaction Response Times
• Throughput
• Availability Requirements
• real users
• Concurrency (application/logged in, application
  active, transaction)

8
Define the Scope of Testing

• Time
• Potential Liklihood of change
• Sufficient Load Generation Capacity
• Just the online application, or also the batch,
  reporting, ETL, middleware, annual processing,
  alternate clients (wireless, time clocks) etc.
• LAN/WAN

9
Test Environment Considerations

• Hardware/Platform
• Size
• Versions
• Load Balancing
• Network Simulation
• Software Infrastructure
• Application Functionality
• Monitoring Tools

10
Other Considerations

• Usage Patterns
• Transaction Mix
• Load Generation Limitations
• 1000 real users doesnt always require 1000 VUs

11
Test Tracking and Reporting

• Test Parameters (repeatability)
• Test Results
• Throughput
• Response times
• System impacts

12
Sample Test Tracking
13
Sample Results Tracking
14
LDAP Sample Log
15
It Can Get Very Complex
16
Templates Available

1. Transaction Mix Identification
2. Test Plan Creation
3. Load Generator Identification/Installation/Operati
   on
4. Initial/Validation Tests
5. Initial/Validation Report
6. Primary Load/Performance Tests
7. Primary Load/Performance Report

17
Sample Results
18
Conclusions

• Planning Makes All the Difference!!!
• Know Why You Are Testing
• Understand The Test Environment
• Understand The Business/Customer Requirements
• Know Who To Involve
• Test Like The Wind