What design considerations are necessary for designing applications for a clustered deployment? Are there any J2EE design pattern(s) that has a potential to misbehave in a clustered environment? Any tips/links to web resources/insights from personal experience( ) will be greatly appreciated. Cheers, [ December 30, 2002: Message edited by: Ajith Kallambella ]
Open Group Certified Distinguished IT Architect. Open Group Certified Master IT Architect. Sun Certified Architect (SCEA).
This is actually something that I am working on at the moment - I'm writing a J2EE app with a view to further researching into how clustering works and how best to design for clustering. At the moment I have "regular" stateless and stateful session beans which I've deployed into JBoss. If you've not seen the clustering support in JBoss 3 then I strongly recommend taking a look - it's absolutely fantastic and so easy to setup. We even have a nice demo of how JBoss fails over stateful session beans in-flight when one server instance dies. I have seen some good documentation on designing for clusters so I'll try to dig it out. As a starter though, some of the usual EJB stuff applies - don't rely on read/write statics in bean instances, sync'd blocks, etc. Cheers for now Simon