A couple of interesting research papers around failure, found in The Datacenter as a Computer.
Failure Trends in a Large Disk Drive Population (2007)
Out of all failed drives, over 56% of them have no count in any of the four strong SMART signals, namely scan errors, reallocation count, offline reallocation, and probational count. In other words, models based only on those signals can never predict more than half of the failed drives.
Temperature Management in Data Centers: Why Some (Might) Like It Hot (2012)
Based on our study of data spanning more than a dozen data centers at three different organizations, and covering a broad range of reliability issues, we find that the effect of high data center temperatures on system reliability are smaller than often assumed.