From: Mario Weilguni <mweilguni@sime.com>
To: Andreas Kostyrka <andreas@kostyrka.org>
Subject: Re: Performance of count(*)
Date: Thu, 22 Mar 2007 17:20:02 +0100
User-Agent: KMail/1.9.5
Cc: pgsql-performance@postgresql.org,
	"Jonah H. Harris" <jonah.harris@gmail.com>,
	Merlin Moncure <mmoncure@gmail.com>, Andreas Tille <tillea@rki.de>
References: <Pine.LNX.4.62.0703221149210.13747@wr-linux02>
	<200703221536.32575.mweilguni@sime.com>
	<20070322151714.GA11448@andi-lap.la.revver.com>
In-Reply-To: <20070322151714.GA11448@andi-lap.la.revver.com>
MIME-Version: 1.0
Content-Type: text/plain;
  charset="utf-8"
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline
Message-Id: <200703221720.02450.mweilguni@sime.com>

Am Donnerstag, 22. M=C3=A4rz 2007 16:17 schrieb Andreas Kostyrka:
> * Mario Weilguni <mweilguni@sime.com> [070322 15:59]:
> > Am Donnerstag, 22. M=C3=A4rz 2007 15:33 schrieb Jonah H. Harris:
> > > On 3/22/07, Merlin Moncure <mmoncure@gmail.com> wrote:
> > > > As others suggest select count(*) from table is very special case
> > > > which non-mvcc databases can optimize for.
> > >
> > > Well, other MVCC database still do it faster than we do.  However, I
> > > think we'll be able to use the dead space map for speeding this up a
> > > bit wouldn't we?
> >
> > Which MVCC DB do you mean? Just curious...
>
> Well, mysql claims InnoDB to be mvcc ;)

Ok, but last time I tried count(*) with InnoDB tables did take roughly(*) t=
he=20
same time last time I tried - because InnoDB has the same problem as postgr=
es=20
and has to do a seqscan too (I think it's mentioned somewhere in their docs=
).

(*) in fact, postgres was faster, but the values were comparable, 40 second=
s=20
vs. 48 seconds=20

Maybe the InnoDB have made some progress here, I tested it with MySQL 5.0.1=
8.