From: Shridhar Daithankar <shridhar@frodo.hserus.net>
To: Heiko Kehlenbrink <Heiko.Kehlenbrink@vermes.fh-oldenburg.de>
To: Performance <pgsql-performance@postgresql.org>
Subject: Re: performance comparission postgresql/ms-sql server
Date: Tue, 06 Apr 2004 12:57:13 +0530
Message-ID: <40725BD1.3050008@frodo.hserus.net> (raw)
In-Reply-To: <56755.195.243.253.146.1081233367.squirrel@webmail.fh-oldenburg.de>
References: <48932.195.243.253.146.1081179099.squirrel@webmail.fh-oldenburg.de>
<40718146.1010506@frodo.hserus.net>
<56755.195.243.253.146.1081233367.squirrel@webmail.fh-oldenburg.de>
Heiko Kehlenbrink wrote:
> hkehlenbrink@lin0493l:~> psql -d test -c 'explain analyse select avg(dist)
> from massive2 where dist > (1000000*sqrt(3.0))::float8 and dist <
> (1500000*sqrt(3.0))::float8;'
> NOTICE: QUERY PLAN:
>
> Aggregate (cost=14884.61..14884.61 rows=1 width=8) (actual
> time=3133.24..3133.24 rows=1 loops=1)
> -> Index Scan using massive2_dist on massive2 (cost=0.00..13648.17
> rows=494573 width=8) (actual time=0.11..2061.38 rows=499999 loops=1)
> Total runtime: 3133.79 msec
>
> EXPLAIN
>
> seems to me that most time was needed for the index scanning...
Hmm... I would suggest if you are testing, you should try 7.4.2. 7.4 has some
good optimisation for hash agregates though I am not sure if it apply to averaging.
Also try forcing a seq. scan by turning off index scan. I guess index scan for
so many rows is not exactly good thing even if tuple size if pretty small.
Shridhar
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Reply to all the recipients using the --to and --cc options:
reply via email
To: pgsql-performance@postgresql.org
Cc: shridhar@frodo.hserus.net, Heiko.Kehlenbrink@vermes.fh-oldenburg.de
Subject: Re: performance comparission postgresql/ms-sql server
In-Reply-To: <40725BD1.3050008@frodo.hserus.net>
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox