Re: reasons to separate indexes and data in different tablespaces

From: Keith Moore <kmoore_at_zephyrus.com>
Date: Fri, 16 May 2008 17:45:19 -0500 (CDT)
Message-ID: <4254.206.227.160.10.1210977919.squirrel@lady.zephyrus.com>


Recoverability was also the reason for putting redo logs on a separate disks (mirrored by Oracle), archived logs on a separate disk and contols files on 3 (or at least 2) separate disks. Without Raid, a single disk failure caused a recovery situation and if your archive logs were on the same disk as data, you were toast.

The 28(?? or was it 22) disk layout came from Keven Loney's Oracle 7 Database Administration book. That's the one I first used to learn Oracle so I remember it well. That was the optimum that no one could really use and then there was a 17 disk layout and on down to a minimum of 5 disks I think.

Keith

> There are actually several very good reasons to separate them. The first
> and most important is recovery. If you lose an tablespace with only
> indexes, you can always re-create the indexes. If you lose a tablespace
> with both, you have a recovery involved.
>
> The next is manageability. Its just easier to manage things if you keep
> tables and indexes separate.
>
> The third reason is you may not know or have control of the underlying
> storage, but you should have control of the tablespace layouts.
>

--
http://www.freelists.org/webpage/oracle-l
Received on Fri May 16 2008 - 17:45:19 CDT

Original text of this message