Re: test / live environment, major performance difference

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: Christo Du Preez <christo(at)mecola(dot)com>
Cc: pgsql-performance(at)postgresql(dot)org
Subject: Re: test / live environment, major performance difference
Date: 2007-06-12 18:42:57
Message-ID: 21855.1181673777@sss.pgh.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-performance

Christo Du Preez <christo(at)mecola(dot)com> writes:
> Fast:
> "public";"layertype";"parentid";0.98797;4;2;"{4,1}";"{0.00902256,0.00300752}";"";-0.142857

> Slow:
> "public";"layertype";"parentid";0.00745157;4;7;"{300}";"{0.976155}";"{1,1,4,5,8,12}";0.92262

Well, those statistics are almost completely different, and what the
slow one says is that parentid = 300 accounts for 97% of the table.
So that's why you get different plans. If that is not reflective of
reality, then you have not ANALYZEd the table lately.

Maybe it's a pilot-error problem, like not doing the ANALYZE as a user
with sufficient privileges? IIRC you have to be table owner, database
owner, or superuser to ANALYZE.

regards, tom lane

In response to

Browse pgsql-performance by date

  From Date Subject
Next Message Francisco Reyes 2007-06-13 00:33:03 Re: Best use of second controller with faster disks?
Previous Message Steven Flatt 2007-06-12 18:31:01 Re: performance drop on 8.2.4, reverting to 8.1.4