Re: index v. seqscan for certain values - Mailing list pgsql-performance

From Bruno Wolff III
Subject Re: index v. seqscan for certain values
Date
Msg-id 20040412195552.GA24133@wolff.to
Whole thread Raw
In response to Re: index v. seqscan for certain values  ("Jeremy Dunn" <jdunn@autorevenue.com>)
List pgsql-performance
On Mon, Apr 12, 2004 at 15:05:02 -0400,
  Jeremy Dunn <jdunn@autorevenue.com> wrote:
>
> Agreed.  However, given that count(*) is a question that can be answered
> _solely_ using the index (without reference to the actual data blocks),
> I'd expect that the break-even point would be considerably higher than
> the < 3% (~38,000 / ~1.3M) I'm currently getting.  Does PG not use
> solely the index in this situation??

That isn't true. In order to check visibility you need to look at the
data blocks.

pgsql-performance by date:

Previous
From: "Jeremy Dunn"
Date:
Subject: Re: index v. seqscan for certain values
Next
From: Tom Lane
Date:
Subject: Re: Re: Index Backward Scan fast / Index Scan slow ! (Modifié par Pailloncy Jean-Gérard)