[jxf]

I don't think that's the most common cause. Deleting the production database is obviously disastrous but there are many other reasons for extended downtime, and they're usually about the distributed-ness of the underlying architecture.