Hi!

Problem:

I have an algorithm works on mathematical data structure (some kind of graph). Data structure is quite big (over 1GB), that is why first i need to load data into memory (it takes a while) and set some conditions for start and end. Generally it is all about tasks and resources.

What I need?

I want to use this algorithm inside JBoss (server-side component) probably best solution will be MDB(if you think I not right it will be great to hear your comment). I don't want to load data into memory every single time. For me best situation is something like shared memory over cluster scope (application scope?). Data in memory is read only and it is very static (it will change twice per year maybe). It will be great to have my bean running in cluster (performance and failover support) with cached data to make it faster. Now I think that if it will be possible to store my data in "cluster memory" best solution for me will be Stateless Session Beans managed by container. Please give some notes to help me find the best way :)

Thanks for supporting!