Date: Fri, 26 Oct 2012 10:30:49 -0500
From: "ktm@rice.edu" <ktm@rice.edu>
To: =?iso-8859-1?Q?B=F6ckler?= Andreas <andy@boeckler.org>
Cc: pgsql-performance@postgresql.org
Subject: Re: Query-Planer from 6seconds TO DAYS
Message-ID: <20121026153049.GC2872@aart.rice.edu>
References: <20121025182226.306900@gmx.com>
	<A5F9B84E-0EEB-4D8F-A107-C4D5AB36A4EA@boeckler.org>
	<20121026145513.GB2872@aart.rice.edu>
	<4818E807-E940-44F2-B10E-75038A82945C@boeckler.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
In-Reply-To: <4818E807-E940-44F2-B10E-75038A82945C@boeckler.org>
User-Agent: Mutt/1.5.20 (2009-12-10)
Content-Transfer-Encoding: quoted-printable

On Fri, Oct 26, 2012 at 05:15:05PM +0200, B=F6ckler Andreas wrote:
> Hi Ken,
>=20
> Am 26.10.2012 um 16:55 schrieb ktm@rice.edu:
>=20
> > Hi Andy,
> >=20
> > You have the sequential_page_cost =3D 1 which is better than or equal=
 to
> > the random_page_cost in all of your examples.
> > It sounds like you need
> > a sequential_page_cost of 5, 10, 20 or more.
>=20
> You're right it was sequential_page_cost =3D 1 because it's really irre=
levant what I do here:
> set random_page_cost=3D2;
> set seq_page_cost=3D5;
> '2012-05-01' AND '2012-08-30' -> NESTEDLOOP
> '2012-04-01' AND '2012-08-30' -> SEQSCAN
>=20
> a) there will be a point, where things will go bad=20
>  this is like patching up a roof 'till you find the next hole instead o=
f making it right at the beginning of construction process
> b) they high seq costs might be true for that table (partition at 40gb)=
, but not for the rest of the database=20
>  Seqscan-Costs per table would be great.
>=20
> Regards,
>=20
> Andy
>=20

Hi Andy,

You can set them per tablespace. Maybe you could put the appropriate tabl=
es
that need the higher costing on the same one.

Regards,
Ken