Re: Query plan issues - volatile tables

From: Craig James <craig_james(at)emolecules(dot)com>
To: Brian Herlihy <btherl(at)yahoo(dot)com(dot)au>
Cc: Postgresql Performance <pgsql-performance(at)postgresql(dot)org>
Subject: Re: Query plan issues - volatile tables
Date: 2009-06-04 16:04:12
Message-ID: 4A27F07C.3050904@emolecules.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-performance

Brian Herlihy wrote:
> We have a problem with some of our query plans. One of our
>tables is quite volatile, but postgres always uses the last
>statistics snapshot from the last time it was analyzed for query
>planning. Is there a way to tell postgres that it should not
>trust the statistics for this table? Basically we want it to
>assume that there may be 0, 1 or 100,000 entries coming out from
>a query on that table at any time, and that it should not make
>any assumptions.>

I had a similar problem, and just changed my application to do an analyze either just before the query, or just after a major update to the table. Analyze is very fast, almost always a orders of magnitude faster than the time lost to a poor query plan.

Craig

In response to

Browse pgsql-performance by date

  From Date Subject
Next Message Matthew Wakeling 2009-06-04 16:33:14 GiST index performance
Previous Message Robert Haas 2009-06-04 13:16:23 Re: degenerate performance on one server of 3