Bill McColl has an entry today that points to two papers from Google and CMU that studied the reliability of a pool of 100,000 disk drives. I’m not into re-blogging without adding value, but his entry is really interesting and already pithy. The punchline:
Two conclusions that can be drawn are that (a) high price “enterprise” drives have similar replacement rates to commodity drives, and (b) manufacturer’s published figures on the reliability of drives is often wildly out.
Link to the papers at Bill’s site.