Re: Auto performance tuning? - Mailing list pgsql-performance

From Jim C. Nasby
Subject Re: Auto performance tuning?
Date
Msg-id 20060321113506.GG15742@pervasive.com
Whole thread Raw
In response to Auto performance tuning?  (Orion Henry <lab@orangekids.org>)
List pgsql-performance
On Mon, Mar 20, 2006 at 11:12:34AM -0800, Orion Henry wrote:
>
> I have to say I've been really impressed with the quality and diversity
> of tools here to increase performance for PostgreSQL.  But I keep seeing
> a lot of the same basic things repeated again and again.  Has anyone
> looked into a "smart" or auto-adjusting resource manager for postgres?
>
> Consider for instance you set it to aggressively use system resources,
> then it would do things like notice that it needs more work mem after
> profiling a few thousand queries and adds it for you, or that a specific
> index or table should be moved to a different spindle and does it in the
> background, or that query plans keep screwing up on a particular table
> so it knows to up the amount of stastics it keeps on that table.
>
> Is this a crazy idea or something someone's already working on?

Feel free to submit a patch. :)

Seriously, the issue here is that everyone who donates code for
PostgreSQL already knows how to tune it, so they're unlikely to come up
with a tool to do it for them (which is much harder than you might
think).

There is the configurator project on pgFoundry, which is a start in the
right direction. Perhaps at some point a commercial entity might come
out with some kind of automatic tuning tool as well. But I doubt you'll
see anything come out of the core developers.

Also, note that you could probably write such a tool without embedding
it into the backend, so don't let that scare you off. :)
--
Jim C. Nasby, Sr. Engineering Consultant      jnasby@pervasive.com
Pervasive Software      http://pervasive.com    work: 512-231-6117
vcard: http://jim.nasby.net/pervasive.vcf       cell: 512-569-9461

pgsql-performance by date:

Previous
From: "Jim C. Nasby"
Date:
Subject: Re: Migration study, step 1: bulk write performance
Next
From: "Jim C. Nasby"
Date:
Subject: Re: Perfomance test figures