posted 02-04-2002 11:51 AM         

Context: Load Testing of Web application.

Please share your definition(s) of "concurrent use" in respect to load and performance testing. My understanding is concurrent users are those users who, during the same identified period of time, are all exercising one particular web application. This does not indicate any particular simultaneous impact on any particular page or function, simply number of users using the web application in the same (defined) period of time.

Thanks, we have quite a debate going on this definition.

Regards,

posted 02-04-2002 12:21 PM         

Sounds like a scope/semantic issue. If the requirements are vague and only say "The system shall support 20 concurrent users" it generally means 20 users in the entire application. But, as with most ambiguities, I would get it clarified.

If you are trying to come up with a good way to build your tests, then I personally would test both - simultaneous users logged into the application performing different functionality AND simultaneous users logged into the application performing the same functionality.

You could also do risk management and look at the different functionality, to see which functions are most popular, critical, and important. For example, with a shopping cart application, you will hopefully have more purchases than returns, and more inquiries/searches than purchases.

Does that help?

posted 02-04-2002 01:05 PM         

Hi Tim,

Oh I have no doubt or problem as to what concurrent means or how it applies to our testing effort. Problem is the other impacted folks, PMs, BAs, and developers are all mailing back and forth as to "what concurrent means". So I'm just collecting from testing pros what their take on the defintition of concurrent users in testing a web application.

Thanks,

quote:

---

Originally posted by testgeek:
Hey Nik!

Sounds like a scope/semantic issue. If the requirements are vague and only say "The system shall support 20 concurrent users" it generally means 20 users in the entire application. But, as with most ambiguities, I would get it clarified.

If you are trying to come up with a good way to build your tests, then I personally would test both - simultaneous users logged into the application performing different functionality AND simultaneous users logged into the application performing the same functionality.

You could also do risk management and look at the different functionality, to see which functions are most popular, critical, and important. For example, with a shopping cart application, you will hopefully have more purchases than returns, and more inquiries/searches than purchases.

Does that help?

Talk to you later,
Tim Van Tongeren

---

posted 02-04-2002 08:36 PM         

We have to meet an NFR(non-func req) of 2000+ concurrent users but only 30 simultaneous users.

This means that at any point in time (during the load test) we will have 2000+x users executing various business cases (percentages of which have been defined).
This translates into that many test scripts and that many users in the same scenario, all ramping up and executing.

During such load, the number of users who shall be able to start execution (and complete it successfully) of a particular use case/business case in the same time period (defined to the granularity of 3 seconds) are considered to be simultaneous users.
In my test script, this is implemented as a rendezvous point. Like I need to save a record, so just before the web_submit_data, I have a rendezvous for 30 vusers.

posted 02-05-2002 07:05 AM         

------------------
Simple minds, Simple thoughts!

posted 02-06-2002 07:40 AM         

Regards,