For what I call ~critical data~ such as the redo logs, we do have copy
pools...
Any long term storage data is exported to two tape sets...
general backups (like oracle .dbf files) are just a single copy, if one goes
bad, you restore the previous copy and roll it forward with redo logs.
We also do full backups every day.
AND due to the size of most of our data bases, one day's backups take up
many many tapes so a single piece of media failure will only (at most)
impact a single backup cycle (set).

but really with the 3590 media, over the last six years we haven't had
enough media failures to justify any special measures.
(the 3590 tapes have been really really good)

DWight