It's all about the answers!

Ask a question

Reduce BIRT engine impact on overall RTC performance


Vladimir Amelin (70462126) | asked Sep 12 '12, 8:58 a.m.
edited Sep 15 '12, 4:01 p.m.
Hi all,

When several users generate complex BIRT reports at the same time, RTC application performance degrades dramatically. BIRT consumes 99% CPU resources on application server. Is there a way to lower priority for BIRT engine processes\threads? We agree to wait for report generation longer, but to make RTC as responsive as usual. Any recommendations are welcome. 

Regards,
Vladimir

Comments
Kevin Ramer commented Sep 12 '12, 9:10 a.m.

Agreed. I see "postRenderReport" in the server activities page going for hours at a time.


Vladimir Amelin commented Sep 12 '12, 9:18 a.m.

There is also nice enhancement for such "hanging" reports:

https://jazz.net/jazz/web/projects/Jazz%20Foundation#action=com.ibm.team.workitem.viewWorkItem&id=110671

One answer



permanent link
Clement Liu (1.5k54149) | answered Sep 14 '12, 1:01 p.m.
I am the person who submitted that enhancement :)

We had very serous performance issue two years ago. The culprit is we are using custom attributes as report parameters. We have solved the problem since by developing the Data Warehouse extension. 


So I would encourage you to look into your report logic for any CPU-intensive calculation first. For example, your logic might be iterating through the huge DW table such as WORKITEM_STATES, which would surely take hours.

Or you can set up the CLM clustering available in 4.0 to load balance the traffic. 

We're planning on moving to RRDI since it can be run in a seperated run time from RTC JVM but Live RTC data is not yet supported...

Your answer


Register or to post your answer.