Say I have 100 concurrent users at any point of time in system. Now I want to measure the throughput of the system. As all 100 users will not be in action at any point of time as there will be some think time as well.

My question is how many request I assume users must be sending to system at any instant of time as rule of thumb to start with ? If I assume all will be inaction at any point of time, it will 100 RPS(request per second ) but that does not look like real to me.

2 Answers 2


You are approaching this from the wrong side. You should ask yourself what was the system designed for and move on from there.

Talk to the developers, the system architects, sales people and managers, read the design documents, read what your competition are doing and whether there are relevant standards (either official or de facto). Ask for real users performance numbers and even better recording of sessions and logs.

You could be finding performance problems even before the first bit was sent into the air


Service Level Objectives are defined mainly by product people (read more at Google SLO/SLA).

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.