2

Newbie perf test question: I read something a while ago that made me think that minimum number of concurrent users was a useful stat to have (annoyingly I can't remember what the reason was). But I just can't think of a reason why it would be useful. We have the maximum number of concurrent users and we would ramp up to that number. I guess it would be good to ramp up to the minimum number of concurrent users first and run the tests at that level, but we could just as easily run the tests at 50-60-80% capacity before running at 100% capacity.

Does minimum number of concurrent users give us anything useful?

Improve this question

3 Answers 3

Reset to default
1

Maybe this: if your system can't scale to the minimum number of users, it doesn't work.

Improve this answer
0

Depending on performance testing type you're executing.

In the absolute majority of cases it's very important metric as well-behaved load tests assume ramp-up period during which virtual users are arriving.

You might want to analyze the influence of onboarding increasing amount of virtual users and correlate their number with other performance metrics such as:

  • response time
  • transactions per second
  • bytes per second
  • number of errors
  • etc.

So you would be able to state the impact caused by increasing load, identify the saturation point, identify the first bottleneck, etc.

If you release all the virtual users at once and your application breaks - the only information you will have is that it doesn't support X users, but the above metrics will not be available and if you increase the load gradually you will be able to see what response time looked like when there were 20/40/100/1000 online users.

Improve this answer
1
  • 1
    But to ramp up users you would have to start from 0. So if minimum number of concurrent users is 1000 and maximum is 5000, the tests will need to ramp up from 0 - 5000 which will go through 1000. Should the tests be ramping up to 1000 and then running at that level specifically? Or should they just be ramping up to say 20% (1000), 30% 80% etc and running at those levels? If the minimum number of concurrent users is 1, does it make sense to test for just 1 person?
    – rozza
    Jan 8, 2019 at 15:42
0

Having this metric you can map it to the system resource monitor to make sure it consumes reasonably small amount resources (RAM, CPU, etc) so that when you have just published your app (say to the could) and that has quite a small audience (that is planned and wrapped in form of minimal concurrent users requirement), your costs for the cloud-based hosting will be acceptable.

Improve this answer

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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