Re: calculating shared data memory space - Mailing list pgsql-general

From Scott Marlowe
Subject Re: calculating shared data memory space
Date
Msg-id dcc563d10801071200u43d1d2c2gfdb86e41ff0d6f74@mail.gmail.com
Whole thread Raw
In response to calculating shared data memory space  (Jean-Michel Pouré <jm@poure.com>)
List pgsql-general
On Jan 7, 2008 11:29 AM, Jean-Michel Pouré <jm@poure.com> wrote:
> Dear Friends,
>
> I am running a phpBB forum with more than 400.000 messages.
> I would like to make sure that all indexes fit in shared memory.

You're kind of leaning towards tying to optimized things in postgresql
the way other dbs are optimized.

Not to say that increasing shared memory won't help, especially with
something like phpBB. It likely will.

A good setting for shared_buffers on a machine running 8.x and with a
fair bit of memory is usually somewhere in the 25% range.  But that is
just a guideline.  Set it to various higher and lower settings and
test to see which is best for you.

Note that the OS caches data as well, and does so quite efficiently.
The real advantage to larger shared_buffers is in the db being able to
fit what it's currently working on into its own memory in one big
chunkj.

> How can I calculate the needed space of all indexes?
> I remember this was part of VACUUM FULL ANALYSE or the like.

That's the Free Space Map I think you're talking about, and it doesn't
need to be a vacuum full analzye usually, just vacuum analyze.

pgsql-general by date:

Previous
From: Tom Lane
Date:
Subject: Re: WAL Sequence
Next
From: Tom Lane
Date:
Subject: Re: Hash Indexes