First of all: We are using Hibernate since almost 2 years and are very happy with it!

We are now in the need of integrating a Master-Slave replication in our existing PostgreSQL server. For now we are still evaluating possible solutions and want to give SkyTools a try.

Unfortunately SkyTools does not work with tables, consisting of multi column primary keys. We have approx. 5-10 tables that have primary keys with 2 or more columns and are mapped by Hibernate. To get rid of those multi column pks we'd add a new synthetic key (+ sequence) to the affected tables and switch the pk to the new synthetic key.

We rely on the second-level cache a lot and still want to benefit from it, after converting the existing tables. And therefore we'd like to know the following:

1. Does the second-level cache merely rely on the pk, set in the table?
2. If not, can Hibernate be configured to use other columns as the ones defined in the pk?

1.yes second level cache rely on primary key mapped for entity

2. u can specify in mapping natural-id - alternative key in db - and u can retrieved object from 2nd level cache by this natural id like this: http://www.hibernate.org/hib_docs/v3/re ... -naturalid

Thanks for your quick reply!

I didn't know until now that there such a thing like <natural-id> exists. ("Hibernate in Action" does not mention it by any word.)

But to let <natural-id>s work, I'm forced to use criteria syntax, right?

hmm i dont know how to write it in hql or if it is possible