Alex,
I agree that this is something that is worth spending time on. This resembles the Oracle RAC (Real Application Cluster). While other people may feel that the amount of data is unreasonable I have a similar problem that will only be solved using such a solution.
In regards to how your database is designed? Who cares? This is an RFC for a general discussion on how to design this level of functionality into Postgres.
IMHO a general discussion isn't too helpful, you might be discussing stuff that's never needed for PostgreSQL. Different database systems give different solutions to the same problem, as you might see from e.g. table partition discussions, which where initiated by Oracle-originating people.
There still might be weaknesses in pgsql, but to identify them, *real* issues need to be discussed. This is necessary to avoid major hardware/software dbms efforts that might well be replaced by organizational/app level tools.
Regards, Andreas
---------------------------(end of broadcast)---------------------------
TIP 9: the planner will ignore your desire to choose an index scan if your
joining column's datatypes do not match
