After rule number doubled from 15,000 to 30,000 this phase, RMA performance decreased sharply, and BrUnit can not execute from RMA. Websphere server complains about CPU starvation, checked with IBM's doc, found one solution: increase heap size. The heap size was set as ms1536m, mx2048m now and everything works well. The heap size is abnormally high when compare with the other servers, common server with 512m heap size can easily support sereral applications with data base connection, jms connection, current transation. The further concern is for 32-bit JVM, max heap size is 2048m, 1536m is very close to it, there's no room anymore. So far it's not certain how many rules will be added in the next phase, if over 10,000 rules again RMA will die mostlikely. Is there another solution from RMA/Blaze side to resolve CPU starvation error?
Any solution/guide or even guess are appreciate. Please share your experience and acknowledge about this topic with us.