On Wed, 16 Mar 1994 17:10:02 EST Wayne T. Smith said:
>   By the way, depending on the implementation, 4 daily/ weekly/
>   quarterly backups means that ADSM database space is 12 times what it
>   normally is.  And it is normally huge at about 700 bytes per file.  If you
>   do database mirroring (who dares not to?), that would be an incredible
>   700*2*12 or 16,800 bytes of database per file.  God doesn't have that
>   much disk space.  :-(

Not necessarily.  It depends on how this would be implemented.  If the
design requires 1 database record for each of the 12 backups, then what
you say is correct.  It is conceivable that this could be designed such
that one database record per file, with (optional) extensions for each
backup copy, could be used.

If a file hasn't changed over the span of your 12 backups, it should be
possible to have only 1 copy of the backup file on the server, and one
database record (with extensions for each of the 12 backups) to represent
that file.  Much of the information in the database record would be
the same for each of the 12 backups, and there isn't really any need to
duplicate that information (except to make the software algorithm easier,
perhaps).

Now, having said that... Even though it might be possible to not have 12
separate backup copies of a file, you might actually WANT to have 12 copies.
The reason for this is that if you want to restore your disk (or filesystem
or volume or whatever), you don't want to have to mount too many tapes to
recover all of the data.  Having 12 copies of the file helps to minimize
the number of tape mounts required to restore an entire filespace.

..Paul