Although there are a few potential bugs in the algorithms and in the implementations presented, the idea is both simple and efficient.

Thus I've implemented a new version which hopefully solves some of the weaknesses I found and has a better theoretical behavior. It is dedicated to PostgreSQL, but may be adapted to other databases.

It is called pg_comparator, a tool for network and time efficient database table content comparison.

see http://www.coelho.net/pg_comparator/ for the perl implementation.