Re: Why is undo backed with redo for GTT

From: Ana C. Dent <anacedent_at_hotmail.com>
Date: Wed, 05 Mar 2003 02:28:23 GMT
Message-ID: <Xns9334BBECC7719SunnySD@68.6.19.6>

Rene Nyffenegger <rene.nyffenegger_at_gmx.ch> wrote in news:b3t4me$1pkgn1$1_at_ID-82536.news.dfncis.de:

>

>> Hello!
>> 
>> Rollback information needs to be stored, for example for rolling back
>> to a savepoint.
>> Rollback information is stored in rollback segments, which are stored
>> in normal/undo tablespaces.
>> Changes to normal/undo tablespaces are normally logged - so far
>> Oracle hasn't made a separate mechanism for handling DML on GTTs.
>> 
>> That's my understanding. In the future it could change..

>
>
> Tanel,
>
> that's exactly my understanding as well, but the question was more if
> there is any cogent reason why redo is generated. If that could change
> in the future, do we live with a less than optimal solution at the
> moment?
>

Try considering this...

The undo/rollback segments reside in a tablespace, just like tables & indexes. When an INSERT, UPDATE or DELETE occurs, one or more blocks in a table changes, zero or more blocks in an index change, one or more blocks in the undo change AND every change MUST be recorded in the redo logfile for safe keeping. Otherwise they could not be recovered (rolled forward and if uncommitted rolled back) after a crash. Received on Tue Mar 04 2003 - 20:28:23 CST