Date: Wed, 9 Nov 2005 13:44:29 +0100
From: "Steinar H. Gunderson" <sgunderson@bigfoot.com>
To: pgsql-performance@postgresql.org
Subject: Re: Improving performance on multicolumn query
Message-ID: <20051109124429.GA9127@uio.no>
Mail-Followup-To: pgsql-performance@postgresql.org
References: <4371E6A7.50808@web.de>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <4371E6A7.50808@web.de>
User-Agent: Mutt/1.5.11

On Wed, Nov 09, 2005 at 01:08:07PM +0100, Jan Kesten wrote:
> Now my problem: I need really many queries of rows using it's primary
> key and fetching about five different columns but these are quite slow
> (about 10 queries per second and as I have some other databases which
> can have about 300 queries per second I think this is slow):
> 
> transfer=> explain analyse SELECT * FROM test WHERE test_a=9091150001
> AND test_b=1 AND test_c=2 AND test_d=0 AND test_e=0;
> 
>  Index Scan using test_idx on test  (cost=0.00..50.27 rows=1 width=1891)
> (actual time=0.161..0.167 rows=1 loops=1)
>    Index Cond: (test_a = 9091150001::bigint)
>    Filter: ((test_b = 1) AND (test_c = 2) AND (test_d = 0) AND (test_e 0))

You don't post your table definitions (please do), but it looks like test_b,
test_c, test_d and test_e might be bigints? If so, you may want to do
explicit "AND test_b=1::bigint AND test_c=2::bigint" etc. -- 7.4 doesn't
figure this out for you. (8.0 and higher does.)

/* Steinar */
-- 
Homepage: http://www.sesse.net/