I am completely new to Performance testing and JMeter and hence my question may sound silly to some people. We have identified some flows of an application and they are like:- Login, SignUp, Perform Transaction. Basically, we are trying to test our API's performance so we have used HTTP Request Sampler heavily. If I have scripted all these flows in JMeter, how can achieve answers to following
How can we decide the benchmark of this system? There is no one in organisation who can help with numbers right now and we have to identify number of users beyond which our system can crash.
For Example, if we say that 1,00,000 users are expected to visit our website in one hour's time then how can we execute this in JMeter? Should Forever loop be used with 3600 seconds(60 mins) of RampUp OR should I go ahead with Number of Threads as 1,00,000 RampUp ask 3600 and Loop Count as 1? What is the ideal way to test this?
What has been done till now? 1. We use to run above mentioned flows with Loop Count as 1. However, as per my knowledge, it's completely based on how much RampUp time I give and JMeter will decide accordingly how many threads it require in parallel to complete the task. Results were not helpful in our case as there was not much load to system. 2. Then, we changed the approach and tried Loop Count as Forever for some 100 users and ran the test for a duration of 10 minutes. After continuing with such test for sometime, we got higher Standard Deviation in JMeter's Summary Report which was fixed by tuning our DB and applying some indexes. We continued this way but I am still confused whether this can really simulate realistic scenario.
Thanks in advance!
