In my previous post I was rather enthousiastic about the Spring JMS library. However, that was before we did run some performance tests. It seemed that with using JMS Spring our code was at least 10 x slower then before! Although the code may be easier to maintain, this performance issue was unacceptable. For this reason I was browsing the internet and found this post. This nice post explains that there are situations in which using JMSTemplate is not the way to go. And apparently, that was the case in our code!
So after rewriting the code we had back our performance and still used some Spring JMS. This just proves Spring is not the solution to every problem.