Date: Wed, 7 May 2003 09:42:27 -0600 (MDT)
From: "scott.marlowe" <scott.marlowe@ihs.com>
To: Achilleus Mantzios <achill@matrix.gatewaynet.com>
Cc: <pgsql-performance@postgresql.org>
Subject: Re: An unresolved performance problem.
In-Reply-To: <Pine.LNX.4.44.0305071359180.27590-100000@matrix.gatewaynet.com>
Message-ID: <Pine.LNX.4.33.0305070940260.8765-100000@css120.ihs.com>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII

On Wed, 7 May 2003, Achilleus Mantzios wrote:

> 
> Hi, few days ago, i posted some really wierd (at least to me)
> situation (maybe a potentian bug) to the performance and bugs list
> and to some core hacker(s) privately as well,
> and i got no response.
> Moreover i asked for some feedback
> in order to understand/fix the problem myself,
> and again received no response.
> 
> What i asked was pretty simple:
> "1. Is it possible that the absense of statistics make the planer produce 
> better plans
> than in the case of statistcs generated with vacuum 
> analyze/analyze?

One of the common examples of this happening was posted a few weeks back.  
someone was basically doing this:

delete from table;
analyze table;
insert into table (1,000,000 times);

the problem was that the table had fk constraints to another table, and 
the query planner for the inserts (all 1,000,000 of them) assumed it was 
inserting into a mostly empty table, and therefore used seq scans instead 
of index scans.

It's not a bug, not quite a feature, just a corner case.