MIME-Version: 1.0
Content-Type: text/plain;
 charset=UTF-8;
 format=flowed
Content-Transfer-Encoding: 7bit
Date: Tue, 30 Oct 2012 14:45:11 +0100
From: Gabriele Bartolini <Gabriele.Bartolini@2ndQuadrant.it>
To: Vincenzo Melandri <vmelandri@imolinfo.it>
Cc: <pgsql-performance@postgresql.org>
Subject: Re: Seq scan on 10million record table..
 =?UTF-8?Q?why=3F?=
In-Reply-To: 
 <CAHSd9GfstpNZHaW+o34S2joPUL03QYoH8s_dMrhMm-f0B2REVw@mail.gmail.com>
References: 
 <CAHSd9GfstpNZHaW+o34S2joPUL03QYoH8s_dMrhMm-f0B2REVw@mail.gmail.com>
Message-ID: <16c001fcdb181afa554d4e60825e7c35@2ndquadrant.it>
User-Agent: [X]con Webmail/0.4.2

 Hi Vincenzo,

 On Tue, 30 Oct 2012 13:15:10 +0100, Vincenzo Melandri 
 <vmelandri@imolinfo.it> wrote:
> I have indexes on both the key on the big table and the import_id on
> the sequence table.

 Forgive my quick answer, but it might be that the data you are 
 retrieving is scattered throughout the whole table, and the index scan 
 does not kick in (as it is more expensive to perform lots of random 
 fetches rather than a single scan).

 To be able to help you though, I'd need to deeply look at the ETL 
 process - I am afraid you need to use a different approach, involving 
 either queues or partitioning.

 Sorry for not being able to help you more in this case.

 Cheers,
 Gabriele
-- 
  Gabriele Bartolini - 2ndQuadrant Italia
  PostgreSQL Training, Services and Support
  Gabriele.Bartolini@2ndQuadrant.it - www.2ndQuadrant.it