Let’s assume that we have a planned performance testing run with 10,000 users.
In this article I will discuss with you why we should start our execution gradually , from my point of view you should not start the run with the mentioned number of users , 10,000 users couldn’t be the start , for the following reasons.
Most of the times when this is the first run ever , it will fail and may be failed dramatically with very high response time and tons of errors.
When we start the run gradually let’s say we started with the 10% of the load which is 1000 users , we can observe while increasing the load the response time changes from run to another also we can observe the backend metrics (CPU , Memory) changes between the runs.
This will absolutely give you a better understanding how the system under test behaving with the user increase.
3. Cost Benefit analysis
Referring to the previous point , it will be clear which part of the system will be worth investing in , here I mean which is more cost saving (enhancing the software or buy/rent more hardware).
In the following I will suggest a simple execution plan , you can tweak it to match your case.
|#||User %||User Number|
As shown on the previous table i suggest 4 steps to reach 100% of the required load , you can absolutely skip steps or decreasing the number of steps if you observe in your results that the difference in response time is minimal when raising the load.
That’s for today 🙂 , hope you find it useful.
Please share your tips, experience, comments, and questions for further enriching this topic of discussion.