>> >> Hello, >> I am investigating MINA to implement a background server process that >> needs to hold millions of objects in memory. Every minute, a percentage of >> these objects needs to be recalculated (this is why we are keeping them in >> memory using coherence). I know there is a limitation of 1.2 GB limit on JVM >> which may not be enough to hold all my objects. > > There is no such limitation on Java. Your underlying system might limit you > though. Try with a 64 bit processor... > >> So I have to partition my objects across multiple servers. So my question >> is, if I use MINA to write this server, what would be a good approach to >> achieve this? > > Absolutely no idea. Keep in mind that MINA is _just_ a NIO framework. It > does not handle your data whatsoever. It just handles IO.
Agree with Emmanuel, MINA can just be Network layer. To me it seems like an Object Cluster. spanning across multiple machines (if necessary). The cache solution like ehcache, Jboss cache may be a possible solution to the task at hand. I would have taken this route, as these solutions come bundled with network layer, and most of the time they use Multicast at NW layer (though its implementation dependent). Though here you could have plugged in MINA to be used as transport layer You can also look at Terracota, can use for POJO clustering -- thanks ashish Blog: http://www.ashishpaliwal.com/blog My Photo Galleries: http://www.pbase.com/ashishpaliwal
