I mentioned yesterday that relying on a server with no redundancy was keeping me up at night. Our server is about to enter its second year and it has only 1 hard disk with no RAID. The diagram below shows why this is such a big deal:
This chart was created by Nate Anderson in his paper [Experts: No cure in sight for unpredictable hard drive loss](http://arstechnica.com/news.ars/post/20070225-8917.html). The chart is based on data from those who know a lot about data farms: Google.
Needless to say, having an 8 percent chance of a drive failure happen **this year** was simply too great a risk for us to live with. Especially since the cost of a week offline would cover the yearly cost of a new server.
The new server from 1&1 has RAID level 1, meaning each drive is a true mirror of the other. If one dies, the other one takes over and someone at 1&1 gets an email telling them to replace a drive.
Needless to say, I will be sleeping better once the new server is in place!