RAID - Creativyst(R) Glossary of developer terms.

Redundant Array of Inexpensive Disks - A standard for grouping disks together for redundancy and performance. The standard describes multiple configurations from RAID 1 to RAID 5.

RAID 1 describes simple disk mirroring. This has a 2X overhead as two bytes of storage is used for every byte of data stored. If one drive fails, its data can be rebuilt on a new drive by copying the data from the non-failing drive. This can be done 'hot'. That is, if a drive fails, a new one can be installed and rebuilt without shutting down the system or requiring the users of the system to stop.

RAID 2-4 Defined, but generally not used.

RAID 5 The most common RAID configuration. This drive spreads four drives worth of data over five drives in such a way that if any one drive fails, its data can be rebuilt onto a new drive using the redundant information contained on the other four disks.

RAID 5 has a storage efficiency of: ( (Number of Drives - 1) / Number of Drives) which is better than the mirrored configuration. Sometimes an extra hot-spare is kept to reduce the duration of recovery-time window (see below), thereby improving fault tolerance.

. . . . .
The problem

All traditional RAID configurations will fail to recover if two or more disks fail within the recovery time window. The recovery time window is the time it takes to replace the failed drive and then rebuild the data on that drive from the redundant drives.

The larger a single drive is, the longer is takes to recover its data from the redundant data on the non-failing drives, and so the higher the odds that a second drive will fail during this time period.

Also, the time it takes to alert a technician and have him or her replace the failed drive is part of this window. Sometimes an extra hot-spare is included in the array which is used to automatically replace any single disk that fails. This, of course, reduces the storage efficiency of any configuration it is used with but increases fault tolerance by reducing the duration of the recovery window.

A new (though proprietary) RAID algorithm called RAID-n is now available, which resolves this problem with traditional RAID.

	Redundant Array of Inexpensive Disks - A standard for grouping disks together for redundancy and performance. The standard describes multiple configurations from RAID 1 to RAID 5. RAID 1 describes simple disk mirroring. This has a 2X overhead as two bytes of storage is used for every byte of data stored. If one drive fails, its data can be rebuilt on a new drive by copying the data from the non-failing drive. This can be done 'hot'. That is, if a drive fails, a new one can be installed and rebuilt without shutting down the system or requiring the users of the system to stop. RAID 2-4 Defined, but generally not used. RAID 5 The most common RAID configuration. This drive spreads four drives worth of data over five drives in such a way that if any one drive fails, its data can be rebuilt onto a new drive using the redundant information contained on the other four disks. RAID 5 has a storage efficiency of: ( (Number of Drives - 1) / Number of Drives) which is better than the mirrored configuration. Sometimes an extra hot-spare is kept to reduce the duration of recovery-time window (see below), thereby improving fault tolerance. . . . . . The problem All traditional RAID configurations will fail to recover if two or more disks fail within the recovery time window. The recovery time window is the time it takes to replace the failed drive and then rebuild the data on that drive from the redundant drives. The larger a single drive is, the longer is takes to recover its data from the redundant data on the non-failing drives, and so the higher the odds that a second drive will fail during this time period. Also, the time it takes to alert a technician and have him or her replace the failed drive is part of this window. Sometimes an extra hot-spare is included in the array which is used to automatically replace any single disk that fails. This, of course, reduces the storage efficiency of any configuration it is used with but increases fault tolerance by reducing the duration of the recovery window. A new (though proprietary) RAID algorithm called RAID-n is now available, which resolves this problem with traditional RAID.
	RAID-n