Re: Anyway to optimize the optimizer

From: jaromir nemec <jaromir_at_db-nemec.com>
Date: Fri, 4 Mar 2005 00:36:49 +0100
Message-ID: <046601c52049$df7e1250$3c02a8c0@JARAWIN>

Hallo Chris,

> To achieve correct partition pruning you should have a restriction on the
> partition
> key.

> If the actual structure doesn't provide it in a simply way, in my opinion,

> the partition key was wrongly chosen!

I can't completely agree with this argumentation. It is very probable that a fact table has more than one related dimensions and therefore there are more access paths. Only one of them can be chosen as a partition key. So a trade off must be made.

A simple example is a table of events that have an opening and closing date.

Does it mean, if I decide to partition on opening date and some one wont to report on closing date, I commited a design flaw?

Regards,

Jaromir

Original Message ----- From: "Christian Antognini" <Christian.Antognini_at_trivadis.com> To: <Waleed.Khedr_at_FMR.COM> Cc: <oracle-l_at_freelists.org> Sent: Thursday, March 03, 2005 11:51 PM Subject: RE: Anyway to optimize the optimizer

--
http://www.freelists.org/webpage/oracle-l

Received on Thu Mar 03 2005 - 18:40:59 CST