Message-ID: <469D2D59.9030106@ifi.uio.no>
Date: Tue, 17 Jul 2007 22:58:01 +0200
From: Thomas Finneid <tfinneid@student.matnat.uio.no>
User-Agent: Thunderbird 1.5.0.12 (X11/20070604)
MIME-Version: 1.0
To: pgsql-performance@postgresql.org
Subject: Re: insert vs select into performance
References: <469D1AD3.9030004@ifi.uio.no>
	<0D6011B2-D09F-45E8-A5D7-5F84ADD1DC2B@seespotcode.net>
	<5666.1184704280@sss.pgh.pa.us>
In-Reply-To: <5666.1184704280@sss.pgh.pa.us>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit


Tom Lane wrote:
> Michael Glaesemann <grzm@seespotcode.net> writes:
>> It would be helpful if you included the actual queries you're using,  
>> as there are a number of variables:
> 
> Not to mention which PG version he's testing. 

Its pg 8.1, for now, I'll be upgrading to a compile optimised 8.2 when I 
do the real test on the real server.

(its on kubuntu 6.10 running on a Thinkpad T60 with dual core 1.5,GB RAM 
and 100GB SATA, just in case anybody feels that is of any interrest.)


> Since (I think) 8.1,
> SELECT INTO knows that it can substitute one fsync for WAL-logging
> the individual row inserts, since if there's a crash the new table
> will disappear anyway.

I am not sure I understand you correctly here, are you saying that 
SELECT INTO in 8.1 disables WAL logging and uses just a single fsync at 
the end? in that case it means that I could disable WAL as well and 
achieve the same performance, does it not?

regards

thomas