On Wed, 12 Nov 2003, Greg Stark wrote:
> > "Marc G. Fournier" <[EMAIL PROTECTED]> writes: > > > Just as a side note, just doing a straight scan for the records, with no > > SUM()/GROUP BY involved, with the month_trunc() index is still >8k msec: > > One of the other advantages of these aggregate tables was that we could > purge the old data much sooner with much less resistance from the > business. Since the reports were all still available and a lot of ad-hoc > queries could still be done without the raw data anyways. Actually, what I've done is do this at the 'load stage' ... but same concept ... ---------------------------(end of broadcast)--------------------------- TIP 4: Don't 'kill -9' the postmaster