Re: Statistics on a table - Mailing list pgsql-general

From Maksim Likharev
Subject Re: Statistics on a table
Date
Msg-id 56510AAEF435D240958D1CE8C6B1770A016D2DA8@mailc03.aurigin.com
Whole thread Raw
In response to Statistics on a table  ("Maksim Likharev" <mlikharev@aurigin.com>)
List pgsql-general
Tried 100 did not help, it seems like n_distinct grows 10000 per 10
statistic,
in my case to be close to reality I have to put 1000.
Is there any complications of that, I mean growing statistic depth other
then
disk space?


-----Original Message-----
From: Tom Lane [mailto:tgl@sss.pgh.pa.us]
Sent: Tuesday, July 08, 2003 11:33 PM
To: Stephan Szabo
Cc: Maksim Likharev; pgsql-general@postgresql.org
Subject: Re: [GENERAL] Statistics on a table


Stephan Szabo <sszabo@megazone23.bigpanda.com> writes:
>> statistic ( n_distinct in particular) records for one of my columns,
>> greatly under calculated it saying:
>> 49726, but in reality 33409816.
>> is it any way how I can improve that?

> You might see if raising the statistics target with ALTER TABLE ALTER
> COLUMN followed by an analyze helps getting reasonable values any.
Maybe
> try a few thousand?

That seems like a big jump, considering the default is only 10.  Try
setting it to 100 (then re-ANALYZE, then check the plan).  If that
doesn't work, try more.  I'd be interested to see what it takes to
get the estimate closer to reality.

            regards, tom lane

pgsql-general by date:

Previous
From: Tom Lane
Date:
Subject: Re: Restoring a postgres database
Next
From: Dennis Björklund
Date:
Subject: Re: Restoring a postgres database