Anyone have any JMS tuning experiences / lessons learned they would like to share?
I will be testing an application running BEA on Redhat 4.0. We will use JMS. The application will make use of Queues and Topics. We will be using JDBC store implamented on Oracle 9i to ensure persistent messeging.
I will be reading up on JMS tuning but any information is appreciated.
Re: JMS Testing
When I was testing a (open)JMS application about a year ago... the jms server (forget what its called) had the shortcoming of only being single or dual threaded. Not sure if this was a shortcoming of how our engineers applied JMS, or it was a JMS shortcoming perse. This caused a bottleneck under heavy traffic.
There are well-documented means of tuning JMS servers/archectures... some ones that come off the top of my head are turning off the message persistency. This may not be acceptable solution... but might be interesting to do to understand the performance tradeoff that is being made.