CODEWITHSUNDEEP
amazon-web-servicesjmeter
sidd607
sidd607

Reputation: 1409

inconclusive results in jmeter during load testing

I am trying to understand the working of jmeter for load testing for aws cloudsearch. I ran 1500 threads I get average load time as 1500 ms but when I run 15 threads i get 10ms as my average load time. Shouldn't both be equal? Can someone please explain this result.

Upvotes: 0

Views: 83

Answers (1)

RowlandB
RowlandB

Reputation: 573

Shouldn't both be equal?

Not at all!

In fact, that's the point. By adding more users (an extra 1485, in your case), you're making the server work harder. It has more requests to process, each of which take Memory, CPU, Threads in various queues, etc. One of these resources is a bottleneck, meaning that it's stopping your server from running as efficiently as you'd like. For instance, if your CPU is stuck at 100%, adding more memory won't do anything.

There are two things that you can (need to?) do next:

1) Find where the bottleneck occurs- does it happen at 1000 users? 500 users?

2) Find the bottleneck, fix it, and run more users.

The key to either of these is monitoring. You need to know what's going on on your server before you can tell what's going wrong on it.

Upvotes: 1

Related Questions