Hard Drive question

casualsax3 · Post by **casualsax3** » Fri Nov 10, 2006 11:33 pm

But it's an *array* made of *independent* drives.

Deathlike2 · Post by **Deathlike2** » Fri Nov 10, 2006 11:40 pm

casualsax3 wrote:But it's an *array* made of *independent* drives.

Meh.

I believe though that this was made for IDE drives (PATA, and now SATA) for the sole reason that such drives have been cheaper than their SCSI counterparts. Hence, "inexpensive".

AntoineWG · Post by **AntoineWG** » Sat Nov 11, 2006 5:00 am

In an array, the drives are dependant on eachother. Despite them being separate physical entities, they function as a single logical unit. Set up a RAID0 array, then remove one drive and tell me how independant the drives are. Even removing 1 drive from a RAID5 or 10 array will adversely affect performance.

Actually, until recently, RAID was almost exclusively implemented with SCSI drives, which tend to be much more expensive than IDE. IDE implementations tended to be either software driven, or at best RAID 1 or 0+1, but the hardware controllers only became mainstream in the last few years.

casualsax3 · Post by **casualsax3** » Sat Nov 11, 2006 6:18 pm

Either through common sense, or when you take a systems engineering class, one of the first things you pick up is that RAID 0 isn't really RAID, since it's not redundant. RAID 1 is the first true raid.

Anyway, all of them are ARRAY's of fucking INDEPENDENT drives - where independent means not the same physical unit. It's got nothing to do with the array being dependent or independent on a drive, or whatever you're on about.

Anyway, the name was changed some time ago at Berkely from inexpensive to independent because it made more sense. While the drives were inexpensive, the hardware was insanely expensive. Also, RAID 1 and 2 were initially created to boost hard drive I/O.

AntoineWG · Post by **AntoineWG** » Sat Nov 11, 2006 10:05 pm

casualsax3 wrote:Either through common sense, or when you take a systems engineering class, one of the first things you pick up is that RAID 0 isn't really RAID, since it's not redundant. RAID 1 is the first true raid.

Anyway, all of them are ARRAY's of fucking INDEPENDENT drives - where independent means not the same physical unit. It's got nothing to do with the array being dependent or independent on a drive, or whatever you're on about.

Anyway, the name was changed some time ago at Berkely from inexpensive to independent because it made more sense. While the drives were inexpensive, the hardware was insanely expensive. Also, RAID 1 and 2 were initially created to boost hard drive I/O.

RAID 0 may not be true RAID because it offers no redundancy, hence the zero, but virtually all RAID controllers support it. It does enhance performance and redundancy when you nest it with RAID 1 to make RAID 10 or 0+1, though.

Deathlike2 · Post by **Deathlike2** » Sat Nov 11, 2006 10:08 pm

AntoineWG wrote:RAID 0 may not be true RAID because it offers no redundancy, hence the zero, but virtually all RAID controllers support it. It does enhance performance and redundancy when you nest it with RAID 1 to make RAID 10 or 0+1, though.

Isn't there RAID 1+0? (eg. order actually matters here?)

AntoineWG · Post by **AntoineWG** » Sat Nov 11, 2006 10:14 pm

Yes. The order does matter. 10 is striping mirrors, 0+1 is mirroring stripes. In RAID 10, all but 1 disk can fail and the array can keep working. In RAID 0+1, only 1 disk can fail before data is lost.