From: Tom Lane <tgl@sss.pgh.pa.us>
To: Hilary Forbes <hforbes@dmr.co.uk>
Cc: pgsql-sql@postgresql.org
Subject: Re: Record size
Date: Wed, 11 Jun 2003 09:11:50 -0400
Message-ID: <10301.1055337110@sss.pgh.pa.us> (raw)
In-Reply-To: <5.1.0.14.0.20030611092555.02ff5e80@mailserver.ipl.co.uk>
References: <5.1.0.14.0.20030611092555.02ff5e80@mailserver.ipl.co.uk>
Hilary Forbes <hforbes@dmr.co.uk> writes:
> Now the question. Does the number of fields that you have in a record and indeed their size affect performance on a SELECT statement when only a subset of fields is being extracted? So suppose that I have
> f1 integer
> f2 varchar(4)
> f3 varchar(20)
> f4 text
> and f4 contains reams of data. (well eg 4k). If I just want f1 and f2, will the performance of the SELECT statement be affected by the fact that f4 is large in physical size? If so, I would be better having two parallel tables one with fields f1,f2,f3 and one with f1,f4 as most of the time I don't want to read in the f4.
Most of the possible benefit applies automatically, because large values
of f4 will be "toasted" (moved out of line). I don't think it's worth
contorting your table structure for. You might care to run some
experiments to verify that theory, though. (But update first;
experiments against 7.0 don't necessarily prove anything about 7.3 ...)
> As a secondary question, presumably it is better to have a permanently compiled view in the database defined as
> CREATE VIEW myview as SELECT f1,f2,f3 from mytable
> rather than issuing the query each time direct to the underlying table?
There's essentially no difference in performance. Views are not
pre-optimized.
regards, tom lane
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: tgl@sss.pgh.pa.us, hforbes@dmr.co.uk
Subject: Re: Record size
In-Reply-To: <10301.1055337110@sss.pgh.pa.us>
* 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