From: Richard Huxton <dev@archonet.com>
To: Olivier Hubaut <olivier@scmbb.ulb.ac.be>
Cc: pgsql-sql@postgresql.org
Subject: Re: Why are these queries so different in time?
Date: Fri, 29 Jul 2005 16:42:48 +0100
Message-ID: <42EA4E78.90305@archonet.com> (raw)
In-Reply-To: <op.suofx0h694ope3@olivier.amaze.ulb.ac.be>
References: <op.suofx0h694ope3@olivier.amaze.ulb.ac.be>
Olivier Hubaut wrote:
> Hi, I have a question about performance querying a 7.4 database. The
> orginal generated query was
>
> SELECT DISTINCT _compound0.object_id AS "ObjectId"
> FROM
(4 LEFT JOINS then a couple of WHERE conditions on 2 tables)
> This on take a huge time to perform, which may come to a timeout on the
> front-end application that uses the database.
> So, I decided to modify manually the query like this:
>
> SELECT DISTINCT _compound0.object_id AS "ObjectId"
> FROM
(Two lots of 2 x Left-joins, unioned together)
> This should give the same result set, but it's really faster than the
> previous one, more than one thousand time faster.
> Is there a reason for this huge difference of performance?
You're probably processing 1000 more rows in the first example. It's
probably running the LEFT JOIN across all the tables then restricting
the results in the WHERE. As it happens you're throwing away duplicates
with DISTINCT and/or UNION anyway, so you never get to see the results.
Try an EXPLAIN ANALYSE of the first example and see if the rows= parts
indicate very large numbers of rows being processed.
To make it faster I'd remove the LEFT JOINs, since your WHERE conditions
seem to rule out the NULL cases anyway.
--
Richard Huxton
Archonet Ltd
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-sql@postgresql.org
Cc: dev@archonet.com, olivier@scmbb.ulb.ac.be
Subject: Re: Why are these queries so different in time?
In-Reply-To: <42EA4E78.90305@archonet.com>
* 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