After a read, a re-read, and digesting the posts before mine, I've come to the conclusion that the best concept in the discussion is using a pull based method to get the backups from the main server(s) to the backup servers. (Whether the backups are stored on tape or disk). It means the server being backed up doesn't even have a clue (much less any access rights) to where it's backups might be being copied.

I am one who has abandoned tape a long time ago. But we've been pushing backups to another server, I'm going to reverse that.

Thanks for the article, and for provoking some thought!