There are 15 Thread Groups i have created and Each Thread group 100 users and one Http Request(Payment Request). I need to start all 15 threads at a time so what will be the Ramp up period time I should give for each thread group.
0
votes
2 Answers
0
votes
It depends on the type of your test, if you're doing something like Spike Test you should keep the ramp-up period as zero.
For "normal" load testing there is no strict recommendation, the only piece of advice is that you need to increase your load gradually so you would be able to correlate increasing load with the increasing throughput, decreasing response time, number of errors, etc. because if you release your threads at once you will only be able to state that your application doesn't support 1500 concurrent users, but you won't be able to tell how many users your application can serve without performance degradation (so called "saturation point", where is the first bottleneck, when the first error occurs, etc.
See JMeter Ramp-Up - The Ultimate Guide article for more details.
According to JMeter official documentation:
Ramp-up needs to be long enough to avoid too large a work-load at the start of a test, and short enough that the last threads start running before the first ones finish (unless one wants that to happen).
Start with Ramp-up = number of threads and adjust up or down as needed
so if you don't have better ideas go for 100 seconds of ramp-up and ensure that your test lasts long enough to include this ramp-up and the same duration of "plateau" phase followed by eventual ramp-down (this way you will be able to check whether application goes back to normal when the load decreases)
0
votes
My suggestion is also a gradual approach but in this particular way: start with one TG to determine what the shortest RAMP-UP time where the 100 threads can be instantiated with no errors AND where the execution reaches a stable steady-state (see below) for a reasonable amount of time. Let's say this RAMP-UP is 30 secs. Repeat the process by adding one more TG and doubling the RAMP-UP duration (60 secs) for both TGs. Again, validate that the 200 threads can be instantiated with no errors and the steady-state is stable. Repeat the process by adding additional TGs one at a time and calibrating the RAMP-UP time accordingly.
Although the short-term goal is to successfully instantiate all 1500 threads in more or less the shortest amount of time, the main focus should be in achieving a stable steady state in terms of KPIs such as Throughput, Response Time, or Utilization at a given workload (VUsers) level.
