Re: Question on large objects

 I wasn't aware of this, so curious. So you mean to say the undo records for lob's are written to the same lob segment during any dml to the lob column and it's not written to the common undo tablespace? And that is causing lot of space consumption because they are not overridden/expired like normal undo?

Also, what about doing alter table move for reorg rather going for dbms_redefinition which needs large code?

On Wed, 27 Sept, 2023, 3:31 am yudhi s, <learnerdatabase99_at_gmail.com> wrote:

> Hello Listers,
> We have some complaints coming from one of the third party databases which
> holds multiple large objects (CLOB/BLOB) in the underlying tables. And the
> size of those growing rapidly and the appdev team keep doing purging
> activity taking downtime. And also mentioned that the size is getting
> decreased significantly if they are doing the table reorg using
> dbms_redefinition method. Not sure if this is expected behaviour in case of
> tables holding large objects.
>
> Can it be because of the heavy DML, but then is it expected that CLOB/BLOB
> would be fragmented severely as compared to normal data types? Or is it
> because UNDO copies are getting stored in the same lob segment thus
> increasing the size of the LOB segments significantly. And any other
> possible easy way to handle such scenarios?
>
> Regards
> yudhi
>

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