On 10 Jan 2003 at 14:57, Chantal Ackermann wrote:
> I have been running VACUUM ANALYZE at first, then removed ANALYZE in
> hope it would improve the performance, and because I don't think that
> ANALYZE is really necessary. In fact, the update itself doesn't slow
> down. It keeps taking about 0.6 sec or even less time. And I don't think
> ANALYZE speeds VACUUM up on the long run, does it?
Well, actually there are several things that vacuum do. When you run analyze,
it does a subset of those things and thus could be bit faster.
Basically it needs to mark those pages for reuse which update has earlier
used. So this frees up some amount of RAM for postgresql. Besides it updates
statistics too.
I would suggest interlacing smaller updates and vacuum rather than doing big
update and a big vacuum. That should take less time overall..
HTH
Bye
Shridhar
--
QOTD: "This is a one line proof... if we start sufficiently far to the left."