ADSM-L

Re: Backup Sets for Long Term Storage

2002-02-13 16:30:04
Subject: Re: Backup Sets for Long Term Storage
From: Richard Cowen <richard_cowen AT CNT DOT COM>
Date: Wed, 13 Feb 2002 15:20:20 -0600
Just some thoughts:

Generating backupsets requires no client resources.
Backupset currently only cover filesystems, not TDP data.
Backupset only covers active data.
Backupset tapes are tracked in volhistory. (along with the command that
created it.)
Backupset tapes are one-per-node.
Backupset tapes can only be refreshed by restoring to a node, backing-up,
and re-generating.

Exporting a node requires no client resources.
Export node does support TDP data.
Export node can do active, inactive, backup, archive or all.
Export tapes are tracked in volhistory.  (along with the command that
created it.)
Export tapes can be more than one node.
Export tapes can only be refreshed by importing, and re-exporting. (no
client activity)
Export tapes can be dry-run imported with preview=yes.
Export tapes can be imported across O/S platforms, and thus may be more
portable.

Archiving requires client resources (cpu, network, etc.)
Archiving only covers active data.
Most TDP data is type=backup.

Since 7 year archives will not expire very often, the only ways to refresh
them are:
        1) Mix them with backup data (as someone has mentioned.)
        2) Use Copypools for them and re-copy them periodically.
        3) Retrieve and re-archive them.
        4) Use move data periodically on all the tape volumes.

Since technology changes much faster than 7 years, one assumes that any
periodic migration process will result in copying old media (say, DLT) to
new media (say, LTO.)  Hopefully, this applies to version upgrades of TSM as
well.  (Companies that offer migration services will be happy to do this for
you!)

Maybe some future TSM utility/command will support export/backupset
duplication, (I suppose a unix dd command would work if the source and
destination both fit on one physical tape.)

As has been pointed out serveral times, the real question is can you turn
that 7 year-old data into information; that is, will your applications still
run on Windows2007 or AIX 6 and did you keep that old Pentium box to run
them on?  (I bet you will be happy all your database data was saved as flat
CSV files, including the meta information to process them, and that you kept
that DICOM display utility for that old medical image data....)

Richard Cowen
Senior Technical Specialist
CNT
<Prev in Thread] Current Thread [Next in Thread>