Server-managed vs storage-managed backup/recovery?
- From: tim@xxxxxxxxxxxxx
- To: oracle-l@xxxxxxxxxxxxx
- Date: Thu, 24 Jun 2004 10:01:53 -0600 (MDT)
All,
I see a lot of IT shops opting for using the various ?mirror-split?
technologies for performing database backups, as opposed to using
tape-management software (i.e. Veritas, Legato, Tivoli, etc) or RMAN. To make
a long story short, I see this as IT management endorsing the concept of
storage-managed storage as opposed to the concept of server-managed storage.
Not a bad solution, makes a lot of sense in a lot of ways. Lots of good
arguments in favor of storage-managed storage, as long as the services on the
server are simple. However, as the existence of this list attests,
database-based applications tend to be complex and laden with options.
I tend to believe that there is more to the recoverabilt than simply getting
the data to tape (or from tape) as quickly as possible, namely:
- partial database recoveries
- point-in-time recoveries
- not backing up READ ONLY tablespaces as frequently as READ WRITE tablespaces
- detecting corruption in datafiles and archived redo log files
- validating backups, validating restores, validating recoverability
- verifying recoverability
- reporting and analysis on backup and restore performance
All of these are things that RMAN does, but ?mirror-split? techniques make
difficult. Still, for shops that have bought into one or more of the various
?mirror-split? options, I find it difficult to try to argue them away from it.
At least, the arguments are difficult until serious recovery problems have
occurred and the huge range of options offered by Oracle have been constrained
by the use of ?mirror-split? backups.
So, in the spirit of ?join them rather than beat them?, has anyone managed to
combine RMAN with ?split-pair? techniques (i.e. EMC ?TimeFinder? and BCVs,
Veritas ?FlashSnap?, Hitachi pair-splits, etc)?
I?ve thought about using the RMAN ?catalog? command to add DATAFILECOPYs to the
catalog following a ?mirror-split?, but RMAN then tends to try to read those
splits during restore operations. Also, the fact of having created a
DATAFILECOPY using a mirror split means that RMAN has not been able to check
the file for corrupted blocks, an important piece of functionality to lose.
I?m also aware of the PROXY operations in RMAN but have not used them. Anyone
know if this functionality integrates into the ?mirror-split? techniques?
Thanks in advance for any and all information!
-Tim
----------------------------------------------------------------
Please see the official ORACLE-L FAQ: http://www.orafaq.com
----------------------------------------------------------------
To unsubscribe send email to: oracle-l-request@xxxxxxxxxxxxx
put 'unsubscribe' in the subject line.
--
Archives are at //www.freelists.org/archives/oracle-l/
FAQ is at //www.freelists.org/help/fom-serve/cache/1.html
-----------------------------------------------------------------
Other related posts:
