I'm attempting to run a Load Test with Telerik Test Studio. I've recorded a user profile and would like to run it with 100 VUs.
How long does Test Studio wait between starting the user profile with each VU? The "sample rate" seems like maybe it would affect this, but I would like to set it so that it waits 500 ms in between each VU, not 5 whole seconds. Is this possible? 100 VUs should break my site but it's only getting a peak of 150 responses received per second. I am assuming that means 150 HTTP requests responded to (I have 50 HTTP requests in my user profile, so this means it's only doing 3 VUs per second).
How can I get it to send more requests at a time at my server?3 Answers, 1 is accepted
You can explicitly configure the test to start with 100 concurrent users (see the attached screen shot).
Also set the ramp time to 0 so the users can be loaded right away.
Let me know if that works for you.
Regards,
Boyan Boev
Telerik by Progress
Rank 1
That's not what I mean. Let's say my load test is to ping a single web page once: http://example.com. I set up 100 VUs to do this. Are all the requests sent to the server at once? Or is there some delay in between them? Other load testing software I've used allows you to configure how long to wait in between each request, say 500 ms.
There is the "Sampling rate" field, but that only goes down to 5 seconds, not 500 ms. And its description seems to indicate that it means something different. I thought this field would be to do with how long it would wait after the 100 VUs had finished, before sending the 100 VUs again.
Is there any sleep time in between these requests being sent at all? Our tests seem to indicate that there is a large amount of wait in between each request being sent, much more than 500 ms.
Let me please explain how the load testing works.
A single machine is able to handle 8 VUs per CPU core. Let's say your machine has 4 CPU cores and you set 100 VUs with 0 ramp time. In this case Test Studio will send 32 requests simultaneously and the other 68 VUs will be in a queue. After the first 32 VUs finish with the request another 32 VU are sent to the server and the queue is supplemented with another 32 VU so there are always 100 VU until the end of the time.
This is a hardware limitation. In order to use let's say 100 VUs simultaneously you can use 2 or 3 machines to run the load test. You need to install there only the run time edition.
The Sampling rate is a frequency which is used to write the results into the results data base.
The only thing which you can use between two requests is the think time. See that article which describes the details.
Hope that helps.
Regards,
Boyan Boev
Telerik by Progress