Well this was actually under Oracle, but I can extrapolate to Postgres given my experience.
The idea tool for the job is a feature that Postgres has discussed but hasn't implemented yet, "partitioned tables". Under Oracle with partitioned tables we were able to drop entire partitions virtually instantaneously. It also made copying the data out to near-line backups much more efficient than index scanning as well.
I think you can get a similar effect by using inherited tables. Create one "master" table, and then inherit individual "partition" tables from that. Then you can easily create or drop a "partition", while still being able to query the "master" and see all the rows.
HTH,
Joe
---------------------------(end of broadcast)--------------------------- TIP 6: Have you searched our list archives?
http://archives.postgresql.org
