Re: strange row count estimates with conditions on multiple column - Mailing list pgsql-general

From Tomas Vondra
Subject Re: strange row count estimates with conditions on multiple column
Date
Msg-id 4D03CE06.1050504@fuzzy.cz
Whole thread Raw
In response to Re: strange row count estimates with conditions on multiple column  (Tom Lane <tgl@sss.pgh.pa.us>)
Responses Re: strange row count estimates with conditions on multiple column  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-general
Dne 17.11.2010 06:58, Tom Lane napsal(a):
>> BTW I think the default estimate used to be 1000, so it was changed in
>> one of the 8.x releases? Can you point me to the docs? I've even tried
>> to find that in the sources, but unsuccessfully.
>
> It's DEFAULT_RANGE_INEQ_SEL, and AFAIR it hasn't changed in quite a while.
> But I wouldn't be surprised if the behavior of this example changed when
> we boosted the default statistics target.

I've been thinking about this and I think it might be improved. If I
understand the logic corretly, it says 'use half of the histogram bin
size'. But the value

#define DEFAULT_RANGE_INEQ_SEL 0.005

says it's always 0.5%, which is not not true if STATISTICS TARGET is not
100. This could actually yield 10x more precise estimates when the
STATISTICS TARGET is set to 1000.

OK, I know the default value is 100, just thinking about how to improve
the estimates.

Tomas

pgsql-general by date:

Previous
From: Andreas 'ads' Scherbaum
Date:
Subject: PostgreSQL@FOSDEM 2011 - Reminder: call for papers ending soon!
Next
From: zab08
Date:
Subject: SSL ques