I need to ensure that multiple client requests execute in parallel. This is due the the nature of processing may have blocking of many seconds. The requests arrive via socket data from remote devices (not a web app request). I see and understand how JMS can be a solution and I see this interesting approach http://www.devx.com/Java/Article/33694/0/page/1 using EJB timers.
I am concerned about the efficiency of either solution. Any idea which is better? Or another approach????
I have a production application that is working just fine for several years that violates recommended J2EE policy and creates threads. This app is undergoing changes (conversion to EJB3 and more) so now is the time to revamp this. I feel my current solution is very reliable but I want to follow J2EE best practices if that is feasible.