RE: Questions on using RMAN to backup a RAC database with TSM

From: <krish.hariharan_at_quasardb.com>
Date: Wed, 5 Mar 2008 10:40:47 -0700
Message-ID: <002d01c87ee8$0db1fce0$6501a8c0@BHAIRAVIPC01>


Peter

I have dealt with RAC backups in all three configurations (with NetBackup which imposes the same constraint)

  1. Dedicated node for all databases
  2. Dedicated node for a given database
  3. Free for all

The crucial part of this is a way of knowing after the fact, where your backup took place or conversely be agnostic to it.  

a)

With single node backup in a RAC configuration you have to ensure

  1. That node is capable of backing up all the databases
  2. Or if different databases get backed up from different nodes, then a document that captures that
  3. Prevent RMAN connection from failing over to a different node to execute the backup of a database via another instance
  4. The node of choice must have the capacity to execute all the data of a given database(s) consistent with your policy

b)

The same is true of NetBackup (which I am familiar with). You can set up multiple channels are configured to reference the nodes from which backup took place (alternate host restore mechanism) then RMAN will locate the backup piece from the correct channel to execute the recovery (channel fail down if you will). I have tested this and if memory serves me right I think this capability was available starting with 9.0.1  

Regards,

-Krish

Krish Hariharan

President/Executive Architect, Quasar Database Technologies, LLC

http://www.linkedin.com/in/quasardb


From: oracle-l-bounce_at_freelists.org [mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Peter Hitchman
Sent: Wednesday, March 05, 2008 4:39 AM
To: oracle-l
Subject: Questions on using RMAN to backup a RAC database with TSM  

Hi,
I have a new 11G RAC set-up to play with!

We use TSM to do our backups and RMAN. I know that TSM associates backups with the client node from which they came.

So:
a) Where is the best place to run rman from? One of the RAC nodes? Or a third-party "application" server.

b) What sort of TSM issues would we face, if the machine we made the backups from failed, when it comes to restore and recovery? I assume that if we backed up using node1 and then tried a restore of an archivelog file from node2, we would need to make TSM think that node2 was node1 for it to find the backup.

Thanks for any comments.

--

Regards

Pete

--

http://www.freelists.org/webpage/oracle-l Received on Wed Mar 05 2008 - 11:40:47 CST

Original text of this message