i've started using the TreeCache method of jboss-cache. i'm trying to store my report objects into jboss cache. most of my report objects when serialized are about 800k each, others may be more. i will be saving/retreiving hundreds of these objects -- total disk space is about 2gigs (file size from using a less effective caching system that i made).
currently, i am able to put these objects into the cache.... but as soon as i start retreiving some of these objects, i run out of memory. my machine has 1gig of memory and i've told jboss to start with parameters: "-Xms128m -Xmx912m". but my machine eventually starts swapping memory to disk -- and eventually crashes.
is jboss-cache meant to be used for big/many objects like this? if so, why does it eat so much memory?
is there a way to specify how much memory jboss-cache will use -- maybe through a eviction policy? if not, is it in the future plans?
In terms of eviction policy, the memory-based policy is in the roadmap but currently not supported. If your object size is quite uniform, you can specify the node size based or time based eviction policy to keep the heap from running out.