Earning the FAIL in failover

This blows my mind: Involver, sort of a competitor to the far more interesting Mosuki :-), recently posted on their blog:

You may have noticed that our site broke this morning. Apparently the power supply on our database server decided to commit suicide while everyone was asleep.

OMG. How can any social network site hoping for some semblance of success (or real funding) be running their DB on a single machine with no failover. I’m totally floored by this. Even when Mosuki was doing private alpha testing 3½ years ago we had a hot-failover cluster. I wonder if they do backups.