From: Hannes Dorbath <light@theendofthetunnel.de>
Subject: Yet another question on LIMIT performance :/
Date: Mon, 06 Nov 2006 15:13:10 +0100
Organization: Hub.Org Networking Services
Lines: 28
Message-ID: <einfmm$r41$1@news.hub.org>
Mime-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Thunderbird 1.5.0.7 (Windows/20060909)
To: pgsql-performance@postgresql.org

Though I've read recent threads, I'm unsure if any matches my case.

We have 2 tables: revisions and revisions_active. revisions contains 
117707 rows, revisions_active 17827 rows.

DDL: http://hannes.imos.net/ddl.sql.txt

Joining the 2 tables without an additional condition seems ok for me 
(given our outdated hardware): http://hannes.imos.net/query_1.sql.txt

What worries me is the performance when limiting the recordset:
http://hannes.imos.net/query_2.sql.txt

Though it should only have to join a few rows it seems to scan all rows. 
 From experience I thought that adding an ORDER BY on the index columns 
should speed it up. But no effect: http://hannes.imos.net/query_3.sql.txt

I'm on 8.1.5, statistics (ANALYZE) are up to date, the tables have each 
been CLUSTERed by PK, statistic target for the join columns has been set 
to 100 (without any effect).


Thanks in advance!


-- 
Regards,
Hannes Dorbath