I've been trying to use EHCache to improve performance of a readonly web application I've developed with Hibernate. I seem to be able to get the cache to come into play when I explicity do a session.load of a object but what about were I've loaded an instance of object A and it has a reference to an instance of Object B? If I issue a call for A again in another session will it load B as well from the cache. Will the reference to B only be handled by the cache if I use session.load or session.iterate for A. Even with all the classes marked <cache usage="read-only" /> I seem to be getting a lot of database activity. About the same as I has beofre I started. I thought when A was pulled from the cache all of it's references would be intact and no further database activity would occur but that doesn't seemto be the case.

Am I misuderstanding the write up or is there something further I have to do? I've specified the cache provider and added the <cache...> tag to all the classes. I can see in the info output that the plugin loads and a cache is created for each class so I think I have it mostly configured correct.

One differnce between my setup though and the examples is that many of the classes involved in the relationships are actually sub-classes. So the cache tag is on the mapping descriptor for the base class. That should also apply to all the sub-classes declared in that smae descriptor right?

I just noticed this forum post. My application uses a connection provided to it to open the session. Would this also affect a read-only cache. The post only discusses read-write caches which isn't what I'm doing. Maybe I better get more familiar with the cache code. Any advise would be welcome.
http://forum.hibernate.org/viewtopic.php?t=926892&highlight=cache

Cache content of readonly web applications. It is a good way to generate static content to file after data changes (the most simple way is to generate it in filter), looks like OSCache has filter implementation for this use case.