LoadUI high discarded request count
Hi,
I am using LoadUI version 2.1.1.
I am trying to run performance tests on a REST service hosted on Windows Azure from a VM with Intel(R) Xeon(R) CPU @ 2 GHz and 3.75 GB RAM.
The test case has 2 ramp generators setup to run for 2 hrs (ramp duration of 15min, peak duration of 1 hr 30 min) with peak rate at 50.
The ramp generators are connected to 2 Soap UI runners which are basically pointing to same API but with different request headers.
In both the Soap UI runners, I have increased the max Queue size to 10000 and concurrent samples to 1000.
Screenshots are attached for reference.
When it reaches peak load (100 requests/sec), a lot of the requests are getting discarded and the Task Manager in the VM shows CPU usage to be hovering around 100%. So does that mean I have to either upgrade the VM or trim down on the peak rate? More importantly, for the current scenario, does this high usage affect the response times which I am getting in the report? Please suggest.
Also, as you can see in the screenshot, there will be some requests which will stay in running status even when the test crosses the 2 hr mark. I have to manually stop it and abort the currently running requests to get the report. Any insights on this issue also would greatly help.
Regards,
Ankush
I am using LoadUI version 2.1.1.
I am trying to run performance tests on a REST service hosted on Windows Azure from a VM with Intel(R) Xeon(R) CPU @ 2 GHz and 3.75 GB RAM.
The test case has 2 ramp generators setup to run for 2 hrs (ramp duration of 15min, peak duration of 1 hr 30 min) with peak rate at 50.
The ramp generators are connected to 2 Soap UI runners which are basically pointing to same API but with different request headers.
In both the Soap UI runners, I have increased the max Queue size to 10000 and concurrent samples to 1000.
Screenshots are attached for reference.
When it reaches peak load (100 requests/sec), a lot of the requests are getting discarded and the Task Manager in the VM shows CPU usage to be hovering around 100%. So does that mean I have to either upgrade the VM or trim down on the peak rate? More importantly, for the current scenario, does this high usage affect the response times which I am getting in the report? Please suggest.
Also, as you can see in the screenshot, there will be some requests which will stay in running status even when the test crosses the 2 hr mark. I have to manually stop it and abort the currently running requests to get the report. Any insights on this issue also would greatly help.
Regards,
Ankush
