Virtual service response time in performance testing

[Mohammed Sheeraz]

Hi Team,

We have set think time spec of 300ms for the response, during performance testing it works for 20 to 30mins and gradually increases the response time up to 1000ms due to this we are getting around 25% failure rate and increasing mainframe CPU utilization due to high response times.

Could you please help how can we restrict response time to behave static on complete test journey.?

And, any idea what could be the possibilities for such behaviour.


Thanks,
Mohammed Sheeraz

[Premalatha Gadde]

Hi Mohammed,

Have you applied performance tuning properties to the VSE? If not, please look at the knowledge article https://knowledge.broadcom.com/external/article?articleId=48518 and configure the properties in your environment.

I would also recommend you to verify the below.
. On your VSE server, DevTest_Home/tmp folder has any java_pidxxx.hprof files. If you see the HPROF files which indicates VSE might be running out of memory and see failures. As you have 300ms set in your response, internal delay queue might be holding your response and causing the out of memory.
. What is the capacity you are using in your VS?
. Did you allocate more memory to the VSE?
If you are running into out of memory then you can increase the VS capacity and see if you can achieve the expected response time.

Prema

[Mohammed Sheeraz]

Hi Prema, 

Thanks for the article that was so informative.
. What is the capacity you are using in your VS?  Currently running with 8
. Did you allocate more memory to the VSE? Yes, around 16GB for VSE

Not sure what has happened after rebuilding the server, it started giving results as expected.

[Vaibhav Jain]

Hi Sheeraz,

The response time for VSE is dependent on multiple factors. Under high TPS, the response time may vary. Please ensure that the hardware is configured as per your requirement.

P.S. - I assume, you are using the perf license

Thanks

------------------------------
Regards,
Vaibhav Jain
Capgemini
------------------------------

[Mohammed Sheeraz]

Hi Vaibhav,

Thanks for your inputs. Yes, we are having good scalable server on AWS it is c5.4xlarge having 16 CPU's with 32 GB of memory.

not sure, what went wrong post rebuild it started working as expected.

Regards,
Mohammed Sheeraz

[Vaibhav Jain]

Cool, Thanks for the update.

------------------------------
Regards,
Vaibhav Jain
Capgemini
------------------------------