[jboss-dev-forums] [Design of JBoss Portal] - Re: portal clustering problem when using optimistic locking

[prabhat.jha@jboss.com]

I spent this weekend watching Olympics and running tests with new hibernate/jbc jar. Here are the results:

With new jar, results are as good as previous jar upto 3-node which is almost linear taking 1400 users on one node as baseline. In 4-node cluster, results are better and it can handle 5300 users which used to be only 4200. With 5-nodes, result is only a bit better, it handles 3100 which used to be 2800.

Just to make sure there was nothing wrong with 5th server, I used another server and results look similar.

I also verified the new jar with Opt. Lock. because this whole issue started when OL was used (please see the first page of this thread). I don't see any problem with OL and  REPL_SYNC combination. So, I think this new jar is better.

I will take thread dumps with 5 nodes with new jar as Brian had asked for. I will also look into  why org.jboss.portal.core.impl.model.portal.PortalObjectImpl.getChild is not  using cache. 



View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4169772#4169772

Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4169772