Those fellow worker are even more of a reason to separate stuff. Those guys/gals are dangerous!

I'd still argue with you, but you're not making a mistake. I think I'm just more conservative.

My large scale scheme has been:

- backup full once or twice a week to local disk

- back up diff every night to local disk (could be more than once a day depending on requirements)

- back up log to disk every 5-15 minutes, depending.

- After backup (any) completes, immediately copy to remote disk.

- keep 1 full, 1-2 diffs, all logs since earliest diff on local disk.

- Keep same on remote disk

- anything older gets copied to tape, then deleted.

- monitor all jobs, if one fails, alert, try to restart, DBA manually runs if needed, including copy/delete.

- test at least one restore once a week, get something off tape at least once a quarter and verify things.

Might be more cautious with more testing at a bank/hospital.