We don't have 7 year retentions but we do have a process that does a
'move data' on any tape that was last written over a year ago.

David

>>> "Allen S. Rout" <asr AT UFL DOT EDU> 8/16/2006 10:20:10 AM >>>
>> On Tue, 15 Aug 2006 14:34:49 -0500, Troy Frank
<Troy.Frank AT UWMF.WISC DOT EDU> said:


> Where would you draw the line with this?  For monthly snapshots,
> you're saying it works better than archives/backupsets.  Would you
> also extend this to replacing yearly archives that need to be around
> 7 years?  I don't see why not, but I've not spent as much time
> mulling it over as you probably have.


Yes; in fact we're setting about doing just such a thing.

In this environment, I expect a (to us) totally alien concern to be
the most important: Tape reliability. You wrote EOT of tape number
one; You will now not touch that tape for seven years, or it's copy
volumes.  How do you assure yourself that the data is legible?

For live data, we usually have churn in our tape pools; expiration and
reclamation usually cycle through the entire corpus of tapes in a
reasonable timeframe.  Long-term storage of static data blows that
model.

Once framed that way the solution is obvious: when possible, take the
"oldest" tape you've got and MOVE DATA on it.  If you can do this a
few times a month, You will sharply curtail the possibility of a
write-only volume.



- Allen S. Rout