This is a fantastic paper studying the failure rates of a positively enormous number of disks. No summary I could write could fully express how much this should be required reading for anyone responsible for even a small disk array. An Analysis of Data Corruption in the Storage Stack |