I have a one-to-many mapping (i.e. Parent/Child) from 'P' to 'C'. Each C object has a many-to-one to 'D', and the D object is really large (i.e. many columns). When Hibernate inflates P, the select statement for each of the associated C objects include *every* column for it's associated D object.

My problem is that there are relatively few (5 max) D objects, but many C objects that point to them. This results in pulling tons of duplicated D object data since there is ultimately an embedded D object within each returned C object (row).

Could this be a result of my session management stragegy since I allocate a session, inflate P, then close the session and return P?

Any suggestions would be greatly appreciated.

-Mitch

Yes it is.
If you do not need D, then you can proxy it not to load it.
Or you can use the second-level cache
Or use the thread local session pattern to keep you session opened for a larger period of time.