Received: from localhost (unknown [200.46.204.183])
	by postgresql.org (Postfix) with ESMTP id 066132E0067
	for <pgsql-performance-postgresql.org@postgresql.org>;
	Fri, 21 Mar 2008 10:25:51 -0300 (ADT)
Received: from postgresql.org ([200.46.204.71])
	by localhost (mx1.hub.org [200.46.204.183]) (amavisd-maia, port 10024)
	with ESMTP id 29035-03 for
	<pgsql-performance-postgresql.org@postgresql.org>;
	Fri, 21 Mar 2008 10:25:45 -0300 (ADT)
X-Greylist: from auto-whitelisted by SQLgrey-1.7.5
Received: from oxford.xeocode.com (87-127-95-194.no-dns-yet.enta.net
	[87.127.95.194])
	by postgresql.org (Postfix) with ESMTP id 1A5E42E0053
	for <pgsql-performance@postgresql.org>;
	Fri, 21 Mar 2008 10:25:45 -0300 (ADT)
Received: from localhost ([127.0.0.1] helo=oxford.xeocode.com)
	by oxford.xeocode.com with esmtp (Exim 4.69)
	(envelope-from <stark@enterprisedb.com>)
	id 1JchFf-0007MI-8I; Fri, 21 Mar 2008 13:25:43 +0000
To: "Laurent Raufaste" <analogue@glop.org>
Cc: <pgsql-performance@postgresql.org>
Subject: Re: PG writes a lot to the disk
In-Reply-To: <669dc9710803210349x764b2ec7k5aa95e33afefa926@mail.gmail.com>
	(Laurent Raufaste's message of "Fri\,
	21 Mar 2008 11\:49\:14 +0100")
User-Agent: Gnus/5.11 (Gnus v5.11) Emacs/22.1 (gnu/linux)
X-Draft-From: ("nnimap+mail01.enterprisedb.com:INBOX.performance" 4878)
References: <669dc9710803190418g45002e87k19cb742579c1eeac@mail.gmail.com>
	<669dc9710803200348x64122249jd2ac5924d78a6da0@mail.gmail.com>
	<D960CB61B694CF459DCFB4B0128514C201E66F9E@exadv11.host.magwien.gv.at>
	<20080320105833.c539476d.wmoran@collaborativefusion.com>
	<10279.1206026439@sss.pgh.pa.us>
	<669dc9710803210349x764b2ec7k5aa95e33afefa926@mail.gmail.com>
From: Gregory Stark <stark@enterprisedb.com>
Organization: EnterpriseDB
Date: Fri, 21 Mar 2008 13:25:43 +0000
Message-ID: <87prtoqkdk.fsf@oxford.xeocode.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
X-Virus-Scanned: Maia Mailguard 1.0.1
X-Archive-Number: 200803/421
X-Sequence-Number: 29375

"Laurent Raufaste" <analogue@glop.org> writes:

> All this make me think that PG was setting some bit on every row it
> used, which caused this massive write activity (3MB/s) in the table
> files. I'm talking about approx. 50 SELECT per second for a single
> server.

Well that's true it does. But only once per row. So analyze would have set the
bit on every row. You could do the same thing with something ligter like
"select count(*) from <table>".

Tom's thinking was that you would only expect a high update rate for a short
time until all those bits were set.

Slony's inserts, updates, and deletes count as updates to the table as well.

-- 
  Gregory Stark
  EnterpriseDB          http://www.enterprisedb.com
  Ask me about EnterpriseDB's RemoteDBA services!