RE: Tablespace reorganisation
- From: "Jeremiah Wilton" <jeremiah@xxxxxxxxxxx>
- To: <vikramsingh120@xxxxxxxxx>
- Date: Wed, 31 May 2006 11:48:43 -0700
On Wednesday, May 31, 2006 6:37 AM vikram singh wrote:
> I feel the following Link too would be of help to you. However, please let
> me know.
> http://www.embarcadero.com/resources/tech_papers/rswhitepaper1.pdf
Although the overall message in that paper is sound (you shouldn't
reorganize tablespaces) it whitepaper is full of errors. It claims that
tablespace reorganization is necessary for all kinds of mythical reasons.
> "The reason for performing a reorganization inside Oracle is primarily due
> to ... fragmentation..."
> "... fragmentation ...[is]... an object growing into multiple space
> extents."
> "... the ... expansion of the object into multiple extents can take some
> time for Oracle to perform as extra extents are allocated and recursive
> calls occur to the Oracle data dictionary."
This is a made up problem. Except in extraordinary situations, allocating
an extent does not impact availability.
> "... objects that have grown into hundreds of extents can take a longer
> amount of time to access, especially if the extents are scattered over
> different parts of the physical server disk."
This is the same old chestnut that has been hashed over for a decade. See
the section called "lots of extents are bad" in my paper:
http://www.ora-600.net/articles/misconceptions.pdf.
> "...availability of the database is threatened when an object attempts to
> grow into another space extent, yet is prohibited from doing so because it
> has reached its maximum extent limit..."
This is a job for monitoring, not tablespace reorganization.
Additional errors persist throughout the paper. I don't think it makes
Embarcadero look very good to have this up on their site. My apologies to
the author if this seems harsh, but the errors are glaring and many of the
misconceptions have been long refuted. I hope someone would point it out if
one of my papers had such egregious errors.
--
Jeremiah Wilton
ORA-600 Consulting
http://www.ora-600.net
> On 5/31/06, Yadavalli Aditya <aditya.yadavalli@xxxxxxxxx > wrote:
>>
>> How can we know that a tablespace needs to be reorganised....?
--
//www.freelists.org/webpage/oracle-l
Other related posts:
