Another perspective

Having read through many of the messages in this thread, it's apparent that almost all of you are looking at mechanical reliability as the measure of a hard drive's reliability. There's another side to this story, and that's data reliability. Over the years an enormous amount of R&D, engineering and investment has gone into making drives more data reliable. By this I mean that what you send to the drive to be saved is what you get back when you want to retrieve it. Mechanical hard drive crashes are pretty obvious events. Losing a bit on track 899, sector 18 in the middle of a big file may be impossible for the ordinary user to detect, or may cause the whole file to be unreadable, or it may cause a RAID rebuild. As we got to multi-terabyte sized drives, the basic statistical mathematics of drive errors pretty much assured that every drive would have errors on it. To prevent this kind of drive failure, the electronics and firmware/software that run the disk drives is much more sophisticated than it was 5, 10 or 20 years ago. Built in error checking and handling routines have been beefed up, and in some cases added where none existed. This is particularly true in Enterprise class drives, high speed SAS drives and if anyone still makes them, high speed Fibre Channel drives. Also, a quick note for all the people who asked for information on Enterprise class drives...try using a search engine. There's a lot of info out there, most of it from the drive manufacturers.