If you are looking to imitate what Netbackup does within the parameters
of TSM you can use archiving. Daily incremental backups are performed
retaining 8 versions. Perform weekly archives that are retained for 13
months and finally perform a year end archive that is retain
indefinitley. Sure you will not have a copy of every file that made it
overnight for the last six months, but you will be very close. And if
you are using archiving for business related retention requirements, you
are covered legally. Another alternative is to evaluate groups of users
and categorize them according to their usage requirements. For instance,
we retain accounting personnel's personal home drives longer than
manufacturing users. We retain more verions of CAD operators home drives
than develpoment. Programmers and developers have been required to
submit all source to source safe and only develop on specfic servers.
Prevents source from being lost due to inadequate retention on user home
drives...
 
 We started the same way. Tried to mimic what we were accustom to (son,
father, grandfather) with something that doesn't work the same way
(TSM). For 1 year we ran much the same way you are. When all three of
our L700 jukeboxes were at 95% capacity it was time to do something. I
reviewed our restore requests and then worked with all departments to
define their specific requirements, we were able to reduce our overall
backup data by 85% in the first 18 months. With diligence we have been
able to maintain an active list of all major departments and their
business requirements. 

>From what I have been through, it was nearly impossible to get managers
to agree to backing up less. Until you put a dollar amount to it. Every
manager loves to be associated with saving money.

Good luck,

Duane
-----Original Message-----
From: ADSM: Dist Stor Manager [mailto:ADSM-L AT VM.MARIST DOT EDU] On Behalf Of
Hart, Charles
Sent: Tuesday, December 07, 2004 9:18 AM
To: ADSM-L AT VM.MARIST DOT EDU
Subject: Management Class Retention Question


Don't laugh... We currently save 180 Act / Deleted / Version Exists etc
of our backup files (use to be 400).  Anyways in our Netbackup Env we do
weekly fulls retained for 13months and an annual full with permanent
retention.  I would like to get our TSM ennv closer to our Netbackup
retention as we have 150+ GB TSM DB's.  Needless to say we could save on
TSM DB and tape.  We aloe have a Company Retention policy in place that
dictates specific retentions, which we accommodate with Archives.

Essentially we would like to retain a bit more than our Netbackup Env,
because if you had recover a file more than a year old you could only
get one version (Ie no point in time).  Is there a good middle of the
road basic retention?  I have searched adsm etc but haven't really found
anything. We know 180 visions is insane... but 5 version would be little
too loose.

Thank you again, this is the best Tech List in existence!!!!


Found a great Tivoli Field guide called "Tivoli Field Guide -
Full-Incremental Rotations Using IBM Tivoli Storage Manager" 
http://www-306.ibm.com/software/sysmgmt/products/support/Field_Guides_Te
chnical.html

Here are some examples in the field guide....

Example 1 --- 5-Week Full-Incremental Rotation 
In this case, the policy we are trying to implement in Storage Manager
is one of doing a full once a week and incremental on the other days.
After 4 weeks and 6 days, the administrator will reuse the tape on which
he did his first full backup. 
Translating this rotation to Storage Manager is pretty straightforward.
A management class can be defined with the following parameters: 
Versions Data Exists: Nolimit 
Versions Data Deleted: Nolimit 
Retain Extra Versions: 34 
Retain Only Version: 34 
These settings would ensure that inactive (old) versions of files are
kept for 34 days before they are discarded by the Storage Manager
Server. 

Example 2 --- 5-Week Full-Incremental Rotation with Extra Weekly Full
Taken Off-Site 
In this case, the desired policy to be implemented in Storage Manager is
one of doing a full twice a week and incremental on six days. One full
is taken off-site for disaster recovery purposes. This means that after
4 weeks and 6 days, the administrator will reuse the tape on which he or
she did his first full backup. And after the same amount of time, he or
she will recycle the oldest off-site full. 
Translating this rotation to Storage Manager is pretty straightforward.
A management class can be defined with the following parameters: 
Versions Data Exists: Nolimit 
Versions Data Deleted: Nolimit 6
Retain Extra Versions: 34 
Retain Only Version: 34 
These settings would ensure that inactive (old) versions of files are
kept for 34 days before they are discarded by the Storage Manager
Server. The off-site requirements are handled with a copy storage pool
and the Storage Manager Disaster Recovery Manager. A copy storage pool
will allow for an exact copy of the on-site data to be taken off-site on
a regular basis. There is no need to do additional backups to get this
off-site copy. 
Example 3 --- 2-Week Full-Incremental Rotation Kept On-Site 
5-Weekly Fulls Kept both On-Site and Off-Site 
Here, the administrator is essentially using two different rotational
schemes. The two week on-site rotation will be the one that is primarily
used for recovery. As some extra "insurance", the policy calls for
weekly fulls. They would be used in the event the administrator needs to
recover something older than two weeks and in the event of a disaster
recovery scenario. 
The implementation of this policy in Storage Manager requires two
management classes. The first is used to accommodate the two week
rotation and the second is used to accommodate the 5 week rotation. The
management classes have the following parameters set: 
Two Week Class - NODE1 Five Week Class - NODE2 
Versions Data Exists: Nolimit Versions Data Exists: Nolimit 
Versions Data Deleted: Nolimit Versions Data Deleted: Nolimit 
Retain Extra Versions: 13 Retain Extra Versions: 34 
Retain Only Version: 13 Retain Only Version: 34 
Two separate Storage Manager virtual nodes are established that map to
each of the classes, although each of these Storage Manager nodes
corresponds to the same physical machine. Node1 runs its backup on a
daily basis to the Two Week Class. Node2 only runs its backup on a
weekly basis and sends it to a different storage pool than Node1. A copy
storage pool is then created for the node using the five week class and
the Disaster Recover Manager is used to manage the off-site data.