Dear everyone.
We are facing with terrible performance of custom webcenter space application. I am trying many technology to enhance caching capability but it does not work. I've already tried to implement coherence cache, http server, webcache
The problem is, when I access the space application via myspace.cuongpt.vnet:80, the response with very very slow rendering. I can see browser had done 3 redirect request to starting return content
- first redirect from my OHS config to redirect myspace.cuongpt.vnet to myspace.cuongpt.vnet/webcenter/portal/btc
- the 2nd redirect was done by webcenter to append ?afrLoop=xxxxxx. This redirect take me 7 seconds to complete
- the 3rd redirect is to access the default page of space. From this moment, content is returned from task-flow by retrieve from UCM
- overal time for loading my home-page is around 14 seconds
I cannot understand what happened with the 2nd redirect !? Why it take very long time to complete ?
I am considering to migrate UCM from filesystem (SAN) to database to test and resolve the problem. But if you have any idea, please give me any suggestion.
That is my hardware infrastructure:
- two webtier nodes that run on RHEL 64GB memory
- two nodes with WebCenter (portal, ucm) that run on Sun Cluster Oracle Sparc T4-4
- database is running on Exadata
I think hardware is not the root cause of problem. My developer team tried to replace content presenters by taskflow. That solutions can make a good things on development laptop, but cannot do anything on production environment.
what do I miss ?
Thank for your attentions and sorry for my bad English.
Cuong Pham