Thanks for the links. Unfortunately the two databases  do not share a consistent schema. The point in my assignment is to separate mutable data from immutable data for performance reasons. I had a suggestion from a friend to implement an interceptor to does the hacking seamlessly. The interceptor thing will sure do the job in my case. However, I am database  bound now, and I have the mandate to stay away from unnecessary database talks. Well, I have fixed the problem so far by fixing up the mappings in the original database to use the immutable object, which is generously cached, and single select statements on the mutable data with a short term cache for they are not very volatile.