From: Chris Browne <cbbrowne@acm.org>
Subject: Re: When to do a vacuum for highly active table
Date: Tue, 30 Aug 2005 18:05:38 -0400
Organization: cbbrowne Computing Inc
Lines: 33
Message-ID: <6064tn6pxp.fsf@dba2.int.libertyrms.com>
References: <20050830211356.48138.qmail@web31606.mail.mud.yahoo.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
User-Agent: Gnus/5.1007 (Gnus v5.10.7) XEmacs/21.4.17 (Jumbo Shrimp, linux)
Cancel-Lock: sha1:Mb6tP0jYqjMBqYOk/ujn4/x8Npg=
To: pgsql-performance@postgresql.org

thing@m-bass.com (Markus Benne) writes:
> We have a highly active table that has virtually all
> entries updated every 5 minutes.  Typical size of the
> table is 50,000 entries, and entries have grown fat.
>
> We are currently vaccuming hourly, and towards the end
> of the hour we are seeing degradation, when compared
> to the top of the hour.

You're not vacuuming the table nearly often enough.

You should vacuum this table every five minutes, and possibly more
often than that.

[We have some tables like that, albeit smaller than 50K entries, which
we vacuum once per minute in production...]

> We are thinking of splitting the table in two: the part the updates
> often, and the part the updates infrequently as we suspect that
> record size impacts vacuum.

There's *some* merit to that.

You might discover that there's a "hot spot" that needs to be vacuumed
once per minute.

But it may be simpler to just hit the table with a vacuum once every
few minutes even though some tuples are seldom updated.
-- 
output = reverse("gro.gultn" "@" "enworbbc")
http://cbbrowne.com/info/spreadsheets.html
Signs  of  a  Klingon  Programmer #3:  "By   filing this  TPR you have
challenged the honor of my family. Prepare to die!"