ADSM-L

Re: Server-to-server COPYPOOL, anyone?

2000-11-03 11:15:11
Subject: Re: Server-to-server COPYPOOL, anyone?
From: ben huber <ben AT KELMAN DOT COM>
Date: Fri, 3 Nov 2000 09:13:48 -0700
> >Still, I'd be grateful for further *sizing* and configuration hints.
>
> I would think that sizing for a copypool is simply the same size as your
> primary data?
> What I notice at our site is that we use one server as a pure target server,
> the only nodes defined on this machine are ADSM/TSM servers. The two source
> servers have databases of 58Gb and 40Gb, but the target database is only
> 812Mb!

The difference in sizes is because the target server only stores the reference
to the virtual volume.  It has no idea what is inside each virtual volume
(similar in a way to the relationship between a TSM server and a tape library -
the library knows what tapes are in it, but hasn't a clue what is stored on each
tape, whereas the TSM server tracks each individual file and knows what tape it
is on and exactly where).

The downside to this is that if you lose your primary TSM server, you cannot use
the target TSM server as a "standby server" that you flip over to if you lose
your primary.  Instead you have to bring up a primary server with your primary
TSM database to be able to pull coherent data from the target server.

> > > secondary site, and ideas how to choose the MAXCAPACITY
> > > of virtual volumes (in relation to tape capacity).

As Paul mentions below, since TSM will only write once to a virtual volume,
there really isn't a direct correlation between the capacity of the virtual
volumes and the capacity of the tapes at the target server.  Depending on the
amount of data transferred, most virtual volumes may be full, but there will
always be small ones kicking around.

We use 2 GB virtual volumes, with 10 GB 3590 tapes at the target server.  Even
so, the target server server typically breaks these 2 GB virtual volumes down
into 5 or more sub-volumes for database purposes.

I can't speak to virtual volume reclamation as we use the server to server
copypool to backup a large archive storage pool that won't ever see
reclaimation.


Ben

> We have 10Gb MAXCAPACITY. The server to server process marks new volumes as
> full as soon as it opens them so it uses each volume only once to write to.
> This means that your virtual volumes could be any size between 0.01 and 10
> Gb. On the target we have a disk storage pool for each source server, 106Gb
> and 78Gb
>
> One server has 2537 virtual volumes with an average size of 4,7Gb. The other
> server copied 43,6 Gb to the copy pools today over 100Mb ethernet.
>
> Erm,...don't know what else to say. Feel free to write direct to my email
> address if you have any other questions.
>
> Paul.
> _________________________________________________________________________
> Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.
>
> Share information about yourself, create your own public profile at
> http://profiles.msn.com.
<Prev in Thread] Current Thread [Next in Thread>