Our rule of thumb is:

1. Never overwrite an existing backup file.

2. Always have at least 2 separate, usable backup files on availiable media.

The first was to handle the situation where an ad-hoc backup was run outside of the usual backup cycle, plus it makes it easy to quickly check if a backup file exists. The second was insurance against media failure.

The solution was to timestamp all backup files to give them a unique name. It does make the backup/recovery scripts more complex to write, but the extra effort is worth the peace of mind.