Response Performance Deployed Service

[michel silva]

Hello guys!

I have some services in place and would like to increase the performance in relation to the response time of the services. Searching the application, finding a property that can be changed according to the need for response time, by default or value set to 100 and if necessary an immediate response, just fill in the value immediately with 0. If you leave the value as 0 for an immediate answer, can I have problems? In my current scenario, if you are using some time, it does not support more than 60 requests in a row. By restarting the VSE, you can make it run sequentially up to 700 orders. This can vary from service to service. Is there any other configuration in devtest that can improve this performance?

Documentation - https://techdocs.broadcom.com/content/broadcom/techdocs/us/en/ca-enterprise-software/devops/devtest-solutions/10-4/using/using-ca-service-virtualization/using-devtest-portal-with-ca-service-virtualization/edit-virtual-services/update-a-virtual-service-manually/update-responses.html#concept.dita_9cb45dbb1b7c200383432d35ec8cd576956a4f44_ChangetheThinkTimeofaResponse

DevTest Version 10.4.0.366

Thank you all.

------------------------------
Michel Silva
Capgemini
------------------------------

[Vaibhav Jain]

Hi Michel,

The thinkTime is defiend at step level and could be modified accordingly. To optimize the performance of vs, you can try to delete the redundant steps and additional filters and assertions as well. 

Thanks

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

[michel silva]

Hi Vaibhav Jain

if I modify the default value of "thinkTime" can I have problems in vs? The project model has minimal steps. When we get a high number of requests the VSE stops running.

Thanks for your help.

Michel Silva
Capgemini

[Vaibhav Jain]

Hi Michel,

I assuming you are using the performance vse license. You can reduce the thinktime value, its shouldn't create any problems in the vs as it is just the wait time associated with each step.

Thanks

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

[michel silva]

Hi Vaibhav,

In my tests, with a high number of requests the VSE stops working. Would a server with a more robust configuration solve my problem? When I put thinktime to zero it responds very well, but the processing increases absurdly.

This is the server configuration:

Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Byte Order: Little Endian
CPU(s): 2
On-line CPU(s) list: 0,1
Thread(s) per core: 1
Core(s) per socket: 2
Socket(s): 1
NUMA node(s): 1
Vendor ID: GenuineIntel
CPU family: 6
Model: 63
Model name: Intel(R) Xeon(R) CPU E7-8870 v3 @ 2.10GHz
Stepping: 4
CPU MHz: 2094.750
BogoMIPS: 4189.50
Hypervisor vendor: VMware
Virtualization type: full
L1d cache: 32K
L1i cache: 32K
L2 cache: 256K
L3 cache: 46080K
NUMA node0 CPU(s): 0,1

              total        used          free         shared    buff/cache     available
Mem:  8011460 | 6549256 | 318856   | 354816  |  1143348   |   1028480
Swap: 7340028 | 205124   | 7134904

[Vaibhav Jain]

Hi Michel,

The performance of VSE could be dependent on multiple params. One of them is the machine configuration as well. Do you see any memory error in logs?

Thanks


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

[michel silva]

There is no log with a memory error. Only the processing goes up until the VSE stops working. The virtualized services only work again when you manually restart the VSE.

[Vaibhav Jain]

Hi Michel,

This could be related to the existing server configuration. Following are few options:
--> Try boosting the infra of the VSE server
--> Try reducing the number of services running on vse
--> Functional license supports approx. 10-12 TPS, try using perf license for higher tps
--> Increase the java heap size for vse

Thanks

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

[michel silva]

Sorry to ask that. But I didn't understand what I meant by "perf" license and how it would be possible to increase the stack size from java to vse?

thanks for your time and your help.

[Vaibhav Jain]

By perf license, i meant performance VSE license. As per my knowledge, there are separate licenses for functional Test VSE and Performance Test VSE. Perf VSE does support higher TPS.

To setup the java heap size for VSE, you can use following params:
-Xms<size>   --> Minimum size

-Xmx<size>  --> Max size

Thanks

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

[michel silva]

Sorry for the delay in responding. I was able to configure it as indicated and we are in the testing phase to see if we have improved performance.

Thanks for the help again.

[Vaibhav Jain]

No worries. Thanks for the update.


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

[Karl Miller]

Check out the documentation on VSE sizing. You may get better performance by add more functional VSEs and segregating those services that get lots of calls from others. EX: 30 virtual Services, 3 services get 50% of transactions. Have 3 Functional VSEs with each having 1 heavily used service and 9 lightly used services. Note: using Live Invocation modes will cause the service to wait for the live system to respond regardless of think time scale.

A Functional VSE is limited to ~10 TPS. This equals ~864,000 transaction per day.

System Requirements

• CPU Speed: 2 GHz or faster
• CPU Cores and Memory
  • 1 Processor Core and 2GB RAM for each 256 Virtual Service Threads 
  • 1 thread for each Virtual Service to support functional tests
• Disk Space: 5 GB of free space
• Approximate Limit ~10 TPS for Standard Virtual Service (Web Service, No custom steps, no Match script, Response times of 100 ms or less)