Journalspace, an online blog hosting site, is down for the count. It appears they had a mirrored drive setup, but no backups to tape. I can tell you from experience that simply mirroring drives is not a valid backup solution. It's great having redundancy, and mirroring/RAID can ensure uptime. But simply keeping a real-time mirror of your data isn't enough, as it seems the folks that ran journalspace.com just found out in the most painful way possible.
The one obvious gotcha is what happened to them - your data gets corrupt. If you have a real-time or near real-time mirror of that data, then guess what? Your backup is also corrupt. Point-in-time, whether to tape or disk, is essential for any critical data, and a valid retention plan is a must as well. Just overwriting the previous backup isn't good enough - what if your backup fails to complete one early AM and the disk or server fails before you can re-run?
Sure, losing a day of changes or data if last nights backup fails sucks, but at least you had a valid backup from the day before to use as a restore point. If you are just re-writing over the previous backup, you wouldn't have had anything to restore from and that is the same result as not backing up at all.
Bottom line? Tape sucks, sure. Tape drives and autoloaders can be cranky beasts. Changing tapes sucks, too. But losing your data and potentially putting your company out of business sucks more.