Re: Fragmentation with restore multiple channels

From: Dragutin Jastrebic <orahawk_at_gmail.com>
Date: Tue, 4 Sep 2012 23:41:34 +0200
Message-ID: <CANGCQwkqYUCqxm=AiLBKmzcJW_k1VvHQ_s73dqqiag0qQsb1vw_at_mail.gmail.com>

Hi,
I can give 2 logical reasons why it simply cannot work this way (I mean rman multiple channel restore leading to fragmentation)

1. If this "fragmentation" really happens, does it mean that our database has new data density, new data clustering, perhaps more index levels and so on? If so, one would need do gather statistics immediately after opening the database since old statistics are not accurate anymore . More than that, we could think that after any Rman restore, we might gather new statistics "just in case"

   Oracle suggests to gather the statistics after massive data load, perhaps afterdata reorganisation (rebuild ,shrink etc) and so on

   With data pump & exp/imp utility you have parameters to control statistics

   behaviour while importing etc

   But with Rman, has anyone ever heard of such a practice ?

2) If this was true then the cloning of database with Rman could not be called cloning anymore

   since the new created destination database is different from the source database

   In the case of the physical standby database ,If we have used rman to copy the

   database and this fragmentation took place,the blocks on standby database    are not the exact copy of primary database and the recovery process will not work

   So it simply does not make sense. Rman is just copying blocks while restoring. (I don't

   see neither how differential & cumulative backups can contribue to any kind of fragmentation)

Regards,
Dragutin

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

Received on Tue Sep 04 2012 - 16:41:34 CDT