To: "Adam Rich" <adam.r@sbcglobal.net>
cc: "'Craig A. James'" <cjames@modgraph-usa.com>,
	"'Guy Rouillier'" <guyr-ml1@burntmail.com>,
	"'PostgreSQL Performance'" <pgsql-performance@postgresql.org>
Subject: Re: High update activity, PostgreSQL vs BigDBMS 
In-reply-to: <022d01c732cc$69f0e2a0$6400a8c0@dualcore> 
References: <022d01c732cc$69f0e2a0$6400a8c0@dualcore>
Comments: In-reply-to "Adam Rich" <adam.r@sbcglobal.net>
	message dated "Sun, 07 Jan 2007 20:26:29 -0600"
Date: Sun, 07 Jan 2007 21:47:52 -0500
Message-ID: <19873.1168224472@sss.pgh.pa.us>
From: Tom Lane <tgl@sss.pgh.pa.us>

"Adam Rich" <adam.r@sbcglobal.net> writes:
> I'm using 8.2 and using order by & limit is still faster than MAX()
> even though MAX() now seems to rewrite to an almost identical plan
> internally.

Care to quantify that?  AFAICT any difference is within measurement
noise, at least for the case of separately-issued SQL commands.

> Count(*) still seems to use a full table scan rather than an index scan.

Yup.  Don't hold your breath for something different.  Postgres has made
design choices that make certain cases fast and others slow, and
count(*) is one case that has come out on the short end of the stick.
If that's your most important measure of performance, then indeed you
should select a different database that's made different tradeoffs.

			regards, tom lane