Hello,

I'm currently trying to convince the TSM developers of adding a
different behaviour to the archive functionality.

We intend to use our TSM environment not only for backup
purposes, but also for archival.
Archiving, from our point of view, is a user centric service
that should enable the user (i.e. owner of a particular TSM
node) to access his/her data from an arbitrary system, as long
as a TSM client is available.
From earlier versions of TSM we knew that archive sets written
by UNIX or Windows clients are mutually invisible - not a nice
behavior, but not a problem.
Now we realized that writing an archive set by the Windows TSM
client - this being a unicode client - makes the archive
*inaccessible* by any non-unicode client, i.e. any UNIX client.
By this, a user who might have archived a lot of stuff from
UNIX systems has an easy way to block himself from his data.
On the best case, we then have manual admin action to give back
the data to a user - and such a person won't come a second time!

Perhaps it's necessary to distinguish the "restore" and "retrieve"
operations, to our understanding of these at least: I see that
a restore has to regenerate not only data, but also all filesystem
meta data (ACLs, NTFS permissions etc.)
In contrast to this, due to our user centric view of the archival
functionality an exact filesystem restoration might be less important.

Is there anyone willing to second this request to Tivoli?

Best regards,
Thomas
http://www.rz.rwth-aachen.de/staff/eifert