All right, this approach was discussed many times on this list with only
one exception - you are the first one willing to do it on separate
instance. What others do: 1 node for backups and 1 or more nodes for
different archival requirements (each node can belong to different policy
domain), both stanzas in dsm.sys (UNIX) or both dsm.opt (Win) files point
to same server instead of two. This eliminates the hassle with unnecessary
virtual volumes - when you need to restore a single small file, the
satellite TSM server "mounts" a virtual volume which results on main
server restore from tape of whole file/volume.
What is the rationale which drove you to two instance implementation?

Zlatko Krastev
IT Consultant






Mark Ferraretto <mark.ferraretto AT DB DOT COM>
Sent by: "ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU>
07.08.2003 03:17
Please respond to "ADSM: Dist Stor Manager"


        To:     ADSM-L AT VM.MARIST DOT EDU
        cc:
        Subject:        Re: Policy Domain Question


Bill,

I wanted exactly the same thing.  Unfortunately, implementing it is quite
difficult in TSM.

* You need to archive to a primary storage pool
* then you need to checkout the tapes from that storage pool manually and
send them offsite.  You can't use DRM to keep track of the volumes as DRM
only works for copy pools and db backups
* You need to manually work out which tapes are required for reclamation
(see the query below) and manually request those tapes back onsite for
reclamation as necessary

In the end I gave up on this and I did the following:
* I created a second TSM instance on our TSM server for managing archive
data only
* I use virtual volumes so the 2nd instance stores its data on the main
instance which has sole control of the tape library
* Instead of archiving data, I use incremental backups on the 2nd
instance.  I use a different management class on the 2nd instance (1 year
instead of 1 month)
* because I'm using incremental backups to perform 'archiving' only about
20% of data changes each month and so the 2nd instance uses much less
space that doing a regular monthly archive
* because the space used is so much less (over a year it's about 400%
less) I've created a copy pool for my archive data as well.  I can afford
to 'waste' the extra tapes and I get quite a few advantages namely:
* I can manage the 2nd instance copy pool the same way as I manage the
main copy pool
* DRM takes care of onsite and offsite tapes
* You can do offsite reclamation - just like a regular copy pool
* The operators don't need to learn any new procedures for managing the
archive tapes
* The archive data is always onsite which makes for easier restoring

Finally, here's the query I mentioned above:

tsm: SM040>q scr q_reclaim_preview f=r
/* Script comment */
/* List volumes in stgpool $1 to be reclaimed  */
/* if threshold was set to $2%                 */
/* ============================================*/
/* Script Name: Q_RECLAIM_PREVIEW              */
/* Author:      Mark Ferraretto                */
/* Parameter:   storage_pool, threshold        */
/* Example:     run Q_RECLAIM_PREVIEW C3 50    */
/* ============================================*/
select -
        VOLUME_NAME as "$1 pool at $2% reclamation" -
        from VOLUMES -
        where STGPOOL_NAME = upper('$1') -
        and PCT_RECLAIM >= $2 -
        order by VOLUME_NAME


--
Mark Ferraretto
Unix Systems Administrator
Deutsche Bank Hong Kong
w: +852 2203 6362        m: +852 9558 8032        f: +852 2203 6971
mark.ferraretto AT db DOT com



                      Bill_Rosette@PAPA
                      JOHNS.COM                To: ADSM-L AT VM.MARIST DOT EDU
                      Sent by:                 cc:
                      [email protected].        Subject:  Re: Policy Domain
Question
                      EDU


                      08/06/03 10:13 PM
                      Please respond to
                      ADSM-L






I am wanting to set up new Policy Domains, with respective Management
classes, and respective Backup and Archive Copy Groups.  I want my
Archives
to go straight to offsite and not be onsite.  Anybody got ideas and how to
do it?

Thank You,
Bill Rosette
Data Center/IS/Papa Johns International
WWJD





--

This e-mail may contain confidential and/or privileged information. If you
are not the intended recipient (or have received this e-mail in error)
please notify the sender immediately and destroy this e-mail. Any
unauthorized copying, disclosure or distribution of the material in this
e-mail is strictly forbidden.