Re: 32K block size tablespace for indexes
Date: Thu, 29 Jan 2009 08:25:46 -0800
Message-ID: <bf46380901290825o6160bb6dqc2f714bba1f7c2b0_at_mail.gmail.com>
On Wed, Jan 28, 2009 at 11:03 PM, hrishy <hrishys_at_yahoo.co.uk> wrote:
> Hi
>
> I have inherited a database where the indexes alone are in the 32k block
> tablespaces.
>
> Now without any documentation or anything for a hint i am just wundering
> why did the original designers design it like that?
>
> What are the pro's and cons of such a approach ?
>
>
It needs to be pointed out that one of the most vocal proponents
of this practice found huge performance gains by doing this, but
only under limited circumstances.
Oh, and a bug that caused blocks to remain off the freelist, or something like that. I don't remember details.
I do know that when the bug was fixed the test no longer saw noteworthy performance gains.
For all the juicy details, you will need to go do a search on the OTN database forum.
Shouldn't be too hard to find, and will probably take all day to read...
Jared Still
Certifiable Oracle DBA and Part Time Perl Evangelist
-- http://www.freelists.org/webpage/oracle-lReceived on Thu Jan 29 2009 - 10:25:46 CST