1

I have a task for load testing on application by 1,000 concurrent users. we are allowed to run load test on pre-prod environment which have 25% from production environment

what is the correct approach?

  1. Run by 250 concurrent users on pre-prod environment and consider the response time of requests on production is equal to pre-prod environment. (ex. Run 250 users pre-prod Response= 1secs. >> Expect >> production Response = 1 secs with 1000 users)

  2. Run by 1000 concurrent users on pre-prod environment and consider the response time of requests on pre-prod is 4 times of production environment. (ex. Run 1000 users pre-prod Response= 4secs. >> Expect >> production Response = 1 sec with 1000 users)

  3. Run by 250 concurrent users on pre-prod environment and consider the response time of requests on production is 4 times of pre-prod environment. (ex. Run 250 users pre-prod Response= 1 secs. >> Expect >> production Response = 4 sec with 1000 users)

1
  • Computing systems are not linear. The assumptions you have are fragile to the way things scale. What risks are trying to investigate exactly? What baselines do you have? What changes are you dealing with between the baseline and the target system? – João Farias Apr 3 at 14:11
1

You cannot extrapolate the results you get for the scaled down environment to the full-size environment, if the scaled down environment has 25% of the production's hardware capacity it doesn't necessarily mean that you will get 4x times better results if you run the same test on production.

You can test:

  1. Code and Database Interaction
  2. External Service Calls
  3. Load Allocation
  4. Monitoring the Application Server
  5. Run a Soak Test
  6. Run a regression test once you have performance baseline is measured

More information: Performance Testing in a Scaled Down Environment. Part Two: 5 Things You Can Test

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.