RE: Question on diff block sizes in DB

From: Hitchman, Peter <Peter.Hitchman_at_thomson.com>
Date: Sun, 18 Apr 2004 08:42:56 +0100
Message-ID: <8D697B8FCEBB3744B3A1E4AE15E8331E02DA83@der-lon-msg-03.derwent.co.uk>

Hi,
Thanks for the update. I'll try to remember this if I ever get to work on Oracle 9i or higher.

Pete

-----Original Message-----

From: oracle-l-bounce_at_freelists.org [mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Mladen Gogala
Sent: 18 April 2004 08:08
To: oracle-l_at_freelists.org
Subject: Re: Question on diff block sizes in DB

Indexes were cached normally, with all other database objects. The tables in question were 20+GB in size (all Bloomberg data about north american stock exchanges, so called NAMR_EQUITY OUT and PX files, for more then 2 years) and
the 16KB buffer pool of 80MB was used for full table scan. All other indexes,
both bit-mapped and B*Tree were 8k and were properly cached. That way I reaped the benefits of the faster full table scan for the huge NAMR tables and I didn't interfere much with the rest of the objects. This 16k buffer pool was, in essence, a recycle buffer pool with a different blocksize, to facilitate faster full table scans. Indexes were unaffected, and that was the idea. It would be great to use compressed tables, because it would really give the big range scans
a boost, but it would prevent me from adding columns and that would be bad, because our beloved mayor of the New York City sometimes adds new columns to the files.

This email has been scanned by the MessageLabs Email Security System. For more information please visit http://www.messagelabs.com/email

Please see the official ORACLE-L FAQ: http://www.orafaq.com

To unsubscribe send email to: oracle-l-request_at_freelists.org put 'unsubscribe' in the subject line.

--
Archives are at http://www.freelists.org/archives/oracle-l/
FAQ is at http://www.freelists.org/help/fom-serve/cache/1.html


-----------------------------------------------------------------

Received on Sun Apr 18 2004 - 02:41:26 CDT