Consider, we have Front end Angular js application where backend is controlled by a API

Now this application contains around 300 API's.

Whether we want to perform load test for all these API's

How can i plan my test plan? Please provide an suggestion

Also we need to write an scenario based upon the URL's

For example,

  1. How application behaves if login is made by n number of concurrent user
  2. How application behaves if user is created by n number of concurrent user

How these test scenarios has to be planned whether in a separate script(separate jmx file) or else in a separate thread group(same jmx file)

note: Test scenario may increase consider it as an example also hardware based scenarios will be covered

  • 3
    Do you mean 300 distinct endpoints or do you mean the application communicates with 300 APIs, each with its own set of endpoints? The answers to this question will be different depending on that. – Kate Paulk Feb 7 '20 at 14:49
  • We also need to know if these API's are built/maintained by your company, or if these are vendor API's and if you're allowed to do any load/performance testing on them. – Lyndon Vrooman Feb 7 '20 at 14:52
  • @KatePaulk Every API is from same end-point – Mohamed Sulaimaan Sheriff Feb 10 '20 at 5:44
  • @Lyndon vrooman all API's us built/maintained by our company – Mohamed Sulaimaan Sheriff Feb 10 '20 at 5:45
  • First - the API is the application that takes requests and returns responses. The endpoint is the distinct call that's made. So as I read this, you have one API with around 300 endpoints, and it's an internal application. – Kate Paulk Feb 10 '20 at 12:39

Your load test needs to represent real-life application usage.

  1. Create a Thread Group per logical group of users for example there will be:

    • x users which are logging in
    • y administrators which will be managing users and content
    • z authenticated users performing CRUD operations
    • etc.
  2. Distribute the load in the more or less anticipated way

  3. Plan some spikes, i.e. most probably in 9 AM in the morning on working day the majority of logins will happen, after that people will be mostly working, then it might be massive re-login after lunch, etc.
  4. If you don't have expected numbers/percentages from the business you can just go for Stress Testing to wit start with 1 user and gradually increase the load until errors start occurring or response time starts exceeding acceptable thresholds (whatever comes the first), once you find the first bottleneck you can report it.

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.