Has anyone done scalability testing with Spring's message-driven POJOs? I am interested in how they perform against heavy load as compared to MDBs.
We've used message driven POJOs
quite a bit with ActiveMQ
and its very fast.
So long as the JCA container can do pooling of resources well, the JMS provider is the most important differentiator for performance. If you email me I can send you a performance report that shows some startling differences between different JMS providers.
So my recommendation would be, use ActiveMQ :). Then whether you use a J2EE application servers's JCA container or you use message driven POJOs with our JCA container
- it shouldn't really matter much in terms of performance - so go with whichever one you're happier with.