CSI-09 COMMUNICATION TECHNOLOGY FAULT TOLERANCE AUTHOR: V.V. SUBRAHMANYAM
RAID RAID is an acronym for Redundant Array of Independent Disks. A disk array combines the capabilities of a number of small, inexpensive disk drives to exceed the performance of a single, large, expensive disk drive.
LEVELS OF RAID There are six levels of RAID: RAID O RAID 1 RAID 2 RAID 3 RAID 4 RAID 5 Each level supports a different storage layout scheme on the disk drives, from mirroring to parity striping.
NEED OF RAID RAID provides real time data recovery when a hard drive fails, increasing system uptime and network availability while protecting against loss of data. Multiple drives working together also increase system performance.
FAULT TOLERANCE The idea behind the fault-tolerant systems is simple: As the system might eventually fail, installation of redundant subsystems will helps in continue to provide uninterrupted service.
FAULT TOLERANCE AND RAID Fault tolerance is an integral part of RAID theory. For RAID systems to work fault tolerance isn’t an option; it’s necessity. RAID uses intelligent controllers and multiple disk drives to maximize data transfer rates. A more common fault-tolerance approach is data striping, which resurrects the old idea of contiguous data placement.
RAID 0 This is known as disk striping. Data is transferred in parallel across an array of disks. Redundancy is not provided in this level and it does not provide data protection.
RAID 1 This is known as Disk Mirroring. Duplicate contents of one disk are written onto another disk. Provides data protection.
RAID 2 Bit-interleaving data across multiple disks with parity information created using Hamming code. The hamming code detects errors that occur and determine which part is in error. Not used because of high performance penalty.
RAID 3 Data is striped across multiple drives and parity is written to a dedicated drive. This is typically implemented at the BYTE level.
RAID 4 Data is striped across multiple drives and parity is written to a dedicated drive. This is typically implemented at the BLOCK level.
RAID 5 Error correction data is striped at the block level across all the drives in the array. Reads and writes may be performed concurrently.
RAID IMPLEMENTATIONS SOFTWARE BASED RAID HARDWARE BASED SCSI RAID CARD EXTERNAL RAID CARD