> It *IS* more unsafe to use RAID0. And MTBF *IS* additive.
No and no. I designed SCSI controllers and disk subsystems (for the storage division of one of the top computer manufacturers) for years, as well as tested disk subsystems. I know how MTBF is determined. > Actually, > more exactly it is reduced and not increased. If you have 1 drive with > a MTBF of 100000 hours you can expect an error every 100000 hours in > average. That's NOT MTBF. MTBF is FAILURE (that's what the "F" in MTBF is for), not error. It's just like tires on a car, they all wear out at just about the same time. When you roll dice, you have the same chance to roll double 6's each time, even if you just rolled 5 of them in a row. MTBF is a very complex statistical determination, but suffice to say, it is absolutely NOT additive at all, and I assure you that adding a second disk to your system to perform RAID 0 does NOT decrease the reliability of your system measurably (unless some other factor is involved, like cooling or power supply). You stand just as much of a chance of choosing the "failure" drive for your single drive then you do having it as part of your RAID 0 system.