Glossary of Disk Array Terminology
Quick Referance
Peter DeVita's (P.Eng) Paper on RAID Systems and Clustering, Paper written for the IEEE.
Below, is a short list of terms used in RAID systems.
Clustering
Active-Active Servers
Active-Passive Server
RAID: (Redundant Array of Inexpensive Drives) An approach to using many low cost drives as a group to improve performance, yet also provides a degree of redundancy that makes the chance of data loss remote.
Disk Mirroring: Disk Mirroring protects data against hardware failure. In its simplest form, a two disk subsystem would be attached to a host controller. One disk serves as the mirror image of the other. When data is written to it is also written to the other. Both disks will contain exactly the same information. If one fails, the other can supply the user data without problem.
Disk Spanning: Several disks appear as one large disk using this technology. This virtual disk can then store data across disks with ease without the user being concerned about which disk contains data. The subsystem handles this for the user.
Disk Striping: This is the fast way of writing to disk drives. Data is written across disks rather than on the same drive. Segment 1 is written to drive 0, segment 2 is written to drive 1, and so fourth until the last disk is written to. Then the next logical segment is written to drive 0 to repeat the cycle until the write is satisfied.
Duplexing: This refers to the use of two controllers to drive a disk subsystem. Should one of the controllers fail, the other is still available for disk I/O. Software applications can take advantage of both controllers to simultaneously read and write to different drives.
Fault Tolerant: Resistant to failure. For example :A RAID mirrored subsystem is fault tolerant since it can still provide I/O if one disk drive in the mirrored system fails.
Host Adapter: The controller that routes data to and/ or from the CPU.
Hot Fix: Identifies a system in which hot spares are available to replace a failed drive with a hot spare. This is not a physical replacement of the drive. The hot spare is simply loaded with data that was previously on the failed drive and then the operation continues.
Hot Patch: Identifies a system in which hot spares are ready.
Hot Spare: A disk drive that is electrically connected to a CPU system that can take over the operation of a failed disk subsystem. A hot spare is differentiated from a Cold Spare in that a Cold Spare sits on the shelf until a disk drive fails.
Hot Swappable A part that can be removed from a computer safely without turning the power off. Examples are Redundant Power Supplies and disk drives.
ISJC: Intelligent Stripe Job Combination. Improves RAID 5 write performance. Has the ability to convert random into sequential writes.
SLED: Single Large Expensive Drive.
RAID Level 0: Disk Striping: Optimum performance and capacity. Concurrent seeks and parallel access to data brings substantial reduction in disk I/O bottlenecks.
RAID Level 1: Disk Mirroring. 100% Redundancy.
RAID Level 3: Disk striping with bit interleaving and dedicated parity disk. Single even parity bits are stored on a separate ECC disk. Data is striped across all data disks. If a disk fails, the lost data can be recovered by reading from the remaining disks and recomputed using the parity.
RAID Level 5:Disk Striping with distributed parity. All disks contain both data and parity checks. Data is interleaved across multiple disks by block. Parity blocks are interleaved across multiple disks so there will be one parity block per sector row. Can exist with other RAID levels at the same time. (50, 51)
Return to Raid Systems
Return to Commercial Systems and Networks
