Re: Bitmap Scan Pages - Mailing list pgsql-general

From Jim Nasby
Subject Re: Bitmap Scan Pages
Date
Msg-id F752FE4D-D222-475D-8B07-275FEBBE143C@decibel.org
Whole thread Raw
In response to Bitmap Scan Pages  (Listmail <lists@peufeu.com>)
List pgsql-general
On Apr 25, 2007, at 7:36 AM, Listmail wrote:
>     Is there a way to know how many pages were hit by a Bitmap Heap
> scan ?
>
>     For instance :
>
> Bitmap Heap Scan on posts
> (cost=56.71..295.24 rows=2123 width=67)
> (actual time=0.575..1.462 rows=2160 loops=1)
>
>     I'd like to know if it hit 2160 pages (ie I should really run
> CLUSTER) or 50 pages (ie. my table is well clustered, everything
> cool).
>     Since, when it's cached, it's so fast anyway I cant' tell the
> difference, but when it's not cached, it is important.
>
>     This would give an interesting probe for EXPLAIN ANALYZE tuning...

If you don't have anything else running in the database and you've
got stats_block_level on, you could probably determine how many pages
are being run by consulting the appropriate pg_statio_* view.
--
Jim Nasby                                            jim@nasby.net
EnterpriseDB      http://enterprisedb.com      512.569.9461 (cell)



pgsql-general by date:

Previous
From: Jim Nasby
Date:
Subject: Re: postgres on Windows: PAE and max memory
Next
From: Jim Nasby
Date:
Subject: Re: reg: bitmap index implementation