ADSM-L

Re: [ADSM-L] Node replication vs. copy stgpools...

2012-08-08 16:23:38
Subject: Re: [ADSM-L] Node replication vs. copy stgpools...
From: Paul Zarnowski <psz1 AT CORNELL DOT EDU>
To: ADSM-L AT VM.MARIST DOT EDU
Date: Wed, 8 Aug 2012 16:20:29 -0400
I haven't done node replication yet.

My opinion is that node replication and copy stgpools do different things.  
When I first saw node replication, I was thinking I could get rid of my copy 
stgpools (stored in a remotely attached tape library), but then I realized that 
it wasn't that simple.  copy pools provide protection against media failure.  
node replication provides protection against site failure.  It would be nice if 
you could configure node replication in a way that would make copy stgpools 
superfluous, but I don't think you can (at least, not without making some 
compromises).

I am intrigued by the possibility of using node replication to migrate a TSM 
server from one architecture to another.  We have archive data mixed in with 
our backup data, so just starting with fresh backups is a bit of a problem for 
us.  Node replication may provide a way to do this.

..Paul

At 02:35 PM 8/8/2012, Allen S. Rout wrote:
>Howdy.
>
>I'd be interested in hearing anyone who's done node replication
>discussing their experiences.   I'm approaching the jump from v5 to v6,
>and am musing about my configuration to be.
>
>I'm currently doing virtual volume copy pools, which have protected me
>adequately, but replicating everything by node sounds way sexier.
>
>But if I do that, do I have something which can function like a copy
>pool?    I'm guessing that, if I lost a volume at the target location
>and DESTROYed it, I could re-replicate to it.   Is there any way to
>reverse the streams and repopulate a volume on the source server?
>
>I could see something like locking the 'source' node, reversing the
>replication config so you can populate the 'source' from the 'target',
>then turning around and re-pointing the rep config and unlocking the
>source, but that's a little itchy;  it's also complex to do this process
>for every node with occupation on a volume.
>
>
>
>I'm moving architecture _and_ version (AIX to Linux, sigh) so I figure
>since I'll be recreating my whole infrastructure one step at a time,
>it's a GREAT time to rethink everything.
>
>- Allen S. Rout


--
Paul Zarnowski                            Ph: 607-255-4757
CIT Infrastructure / Storage Services     Fx: 607-255-8521
719 Rhodes Hall, Ithaca, NY 14853-3801    Em: psz1 AT cornell DOT edu

<Prev in Thread] Current Thread [Next in Thread>