Hello,

I am writing a reporting application. It involves gathering data from a couple of unrelated tables. One of the tables is a huge one. The datasets fetched from there could also be quite big (we agreed that at least up to 20,000 rows should be handled). This table is populated by other processes, my job is only to read from that table.

I am right now wondering whether using Hibernate for this task is a good idea or not. This is all strictly read-only and caching can't be used either. The benefit of Hibernate would be that I wouldn't have to code the data loading from the ResultSet myself. However isn't it true that Hibernate keeps a copy of the loaded objects in the session? This would mean a lot of extra memory usage. Even with FlushMode.NEVER Hibernate must still be prepared that I call flush() manually?

Any comments would be greatly appreciated.

You can evict() the object from the Session immediately after loading it. However, if your application is not cross-plattform (databases) and you just map the result to some simple report JavaBeans and don't have any complex logic in your application (just displaying the beans), there is no real win here compared to plain JDBC and a direct display of data from resultsets.

I would suggest using something like iBatis DB layer if you just care about bean population. Hibernate is probably overkill in that case.