Can anyone help me in effectively analysing the Response Time Graph after a Test Run in Jmeter? @moolyatesting

[shweta gupta]

[Andrey Pohilko]

Please, add some more details on your test. Maybe share a graph screenshot. I can't help you blindly, there may be a lot of different situations.

And in most of situations single Response Times graph is not enough to understand. You'll need a Transactions Per Second and Active Threads Over times graphs to analyse.

среда, 19 июня 2013 г., 10:08:52 UTC+4 пользователь shweta gupta написал:

[shweta gupta]

Thanks Andrey.

Here i have attached Response Time ,Transactions Per Second and Active Threads Over times graphs.

Please have a look.

The Test Run was for a script for searching data in a website.

And i had added a Synchronizing Timer at The end of the 3rd Transaction i.e after entering the search keyword and clicking on enter.

On Wednesday, June 19, 2013 11:38:52 AM UTC+5:30, shweta gupta wrote:

[Andrey Pohilko]

Much better now, I can see your test.

First of all, it has too short duration. Too short tests can't show you the behavior of the system. I suggest you:

1. increase loop count in thread group or use loop controller to achieve longer test duration, at least 3-5 seconds. Correct workload model rules to correct results.

2. Analyse jp@gc Response Times Over Time graph (not "Response Time Graph"!), it will show you a "per second" values, which will be corresponded to Tran per Second and Active Threads graphs.

Could you make those changes and share graphs again? 

четверг, 20 июня 2013 г., 11:22:49 UTC+4 пользователь shweta gupta написал:

[shweta gupta]

I changed the Loop Count to 3 and added a Loop Controller with 2 iterations.

Here are the new Graphs.

What are your observations?

I wanted to actually analyse the change after adding a synchronization point.

[Andrey Pohilko]

Now it's better, but I suggest you increasing loop count to even more value.

Your Synchronizing Timer results that every sampler inside the controller is synchronized. I guess it is not what you wanted... And it looks a bit strange for generic load test. A real workload will not be synchronized in such way.

To analyse the change just disable the synchronization, repeat test and compare the graphs. The difference should be obvious.

четверг, 20 июня 2013 г., 13:34:29 UTC+4 пользователь shweta gupta написал:

[shweta gupta]

Ok ..I made the Loop to 5 and Loop Controller Iterations to also 5.

Disabled the Synch Timer.The graphs look like these now.

[Andrey Pohilko]

So now you can see that without synchronizer you have less response times. 20x times less... Now it depends on what you wanted to achieve.

But let me ask you one more question: why do you have samplers for all those JS stuff? Usually people use "download embedded resources" for main sampler. Seems you have recorded your test plan, but it needs some manual improvement.

четверг, 20 июня 2013 г., 14:27:55 UTC+4 пользователь shweta gupta написал:

[shweta gupta]

Nice Tip Andrey...i didnt know about this feature at all ! 

After using download embedded resource feature,my scripts look optimized now.

Cheers !

Shweta

--
You received this message because you are subscribed to a topic in the Google Groups "jmeter-plugins" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/jmeter-plugins/RzSPW0E07Cw/unsubscribe.
To unsubscribe from this group and all its topics, send an email to jmeter-plugin...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.