Re: large dataset with write vs read clients - Mailing list pgsql-performance

From Greg Smith
Subject Re: large dataset with write vs read clients
Date
Msg-id 4CAE3FDB.1040909@2ndquadrant.com
Whole thread Raw
In response to large dataset with write vs read clients  (Aaron Turner <synfinatic@gmail.com>)
Responses Re: large dataset with write vs read clients  (Aaron Turner <synfinatic@gmail.com>)
Re: large dataset with write vs read clients  (Florian Weimer <fw@deneb.enyo.de>)
List pgsql-performance
Aaron Turner wrote:
> Are newer PG versions more memory efficient?
>

Moving from PostgreSQL 8.1 to 8.3 or later should make everything you do
happen 2X to 3X faster, before even taking into account that you can
tune the later versions better too.  See
http://suckit.blog.hu/2009/09/29/postgresql_history for a simple
comparison of how much performance jumped on both reads and writes in
the later versions than what you're running.  Memory consumption will on
average decrease too, simply via the fact that queries start and finish
more quickly.  Given an even workload, there will be less of them
running at a time on a newer version to keep up.

Given the size of your database, I'd advise you consider a migration to
a new version ASAP.  8.4 is a nice stable release at this point, that's
the one to consider moving to.  The biggest single problem people
upgrading from 8.1 to 8.3 or later see is related to changes in how data
is cast between text and integer types; 1 doesn't equal '1' anymore is
the quick explanation of that.  See
http://wiki.postgresql.org/wiki/Version_History for links to some notes
on that, as well as other good resources related to upgrading.  This may
require small application changes to deal with.

Even not considering the performance increases, PostgreSQL 8.1 is due to
be dropped from active support potentially as early as next month:
http://wiki.postgresql.org/wiki/PostgreSQL_Release_Support_Policy

Also:  PostgreSQL 8.1.3 has several known bugs that can lead to various
sorts of nasty data corruption.  You should at least consider an
immediate upgrade to the latest release of that version, 8.1.22.  Small
version number increases in PostgreSQL only consist of serious bug
fixes, not feature changes.  See
http://www.postgresql.org/support/versioning for notes about the
project's standard for changes here, and how it feels about the risks of
running versions with known bugs in them vs. upgrading.

--
Greg Smith, 2ndQuadrant US greg@2ndQuadrant.com Baltimore, MD
PostgreSQL Training, Services and Support  www.2ndQuadrant.us
Author, "PostgreSQL 9.0 High Performance"    Pre-ordering at:
https://www.packtpub.com/postgresql-9-0-high-performance/book


pgsql-performance by date:

Previous
From: Robert Haas
Date:
Subject: Re: Odd behaviour with redundant CREATE statement
Next
From: Dave Crooke
Date:
Subject: Re: Odd behaviour with redundant CREATE statement