Re: Limited Shared Buffer Problem - Mailing list pgsql-performance

From Greg Smith
Subject Re: Limited Shared Buffer Problem
Date
Msg-id 4B635EFD.7000403@2ndquadrant.com
Whole thread Raw
In response to Re: Limited Shared Buffer Problem  (Cédric Villemain <cedric.villemain.debian@gmail.com>)
List pgsql-performance
Cédric Villemain wrote:
> AFAIK effective_cache_size is estimation of OS Page Cache + Estimated
> Cache in shared_buffers.
>

Yes, the total value you set is used as is, and should include both
pieces of memory.  The planner doesn't add the shared_buffers value to
the total first for you, as some people might guess it would.

The only thing effective_cache_size is used for is estimating how
expensive an index is likely to be to use, to make decisions like when
to do an index-based scan instead of just scanning the table itself.

--
Greg Smith    2ndQuadrant   Baltimore, MD
PostgreSQL Training, Services and Support
greg@2ndQuadrant.com  www.2ndQuadrant.com


pgsql-performance by date:

Previous
From: Greg Stark
Date:
Subject: Re: Faster CREATE DATABASE by delaying fsync (was 8.4.1 ubuntu karmic slow createdb)
Next
From: Greg Stark
Date:
Subject: Re: Constraint propagating for equal fields