VRAID

-*- Mode: indented-text -*- --- $Id: raid.doc,v 1.2 1998/10/07 23:17:42 vitus Stab $ WHAT IS RAID? RAID (redundant array of inexpensive disks) is the name of some standards to increase the security of the disk subsystem. Those standards were first published by Gibson, Katz and Patterson in 1987. The idea of RAID is to use several disks to build a fast, errorprone subsystem. Usuallay this is done by a special SCSI controller but the possibility of software RAID exists, too. Currently the following RAID levels are commonly available: RAID Description Checksum data disks + chksum disks ------------------------------------------------------------------ - Chaining none 2+ 0 Striping none 2+ - 1 Mirroring none 1+ 1 2 Stripe Set Hamming Code 2+ 1 3 Stripe Set XOR 2+ 1 4 Striping XOR 2+ 1 5 Striping, distributed XOR 2+ 1 Besides these 'official' RAID levels 0-5 may vendor-unique levels exist, for example Mylex RAID 6 (combination of RAID 0 and 1) und 7, Siemens RAID 7 and many more. Every vendor has his standards and most standards are different from standards used by other vendors. RAID 0 - Striping Using two of more disks data is written alternating to all disks. Transfer rate is usually much higher as a single disk. Access time remains the same. RAID 0 is no real RAID as it doesn't provide redundancy. Danger of data loss is greater than a single disk as with a loss of a single disk the whole subsystem is worthless. You don't need a controller to implement RAID 0, this can by done by a piece of software. This isn't a real professional solution but may nevertheless increase data throughput. RAID 1 - Mirroring A simple way to increase data security: write all data to two or more disks. So all disks contains exactly the same data. Disadvantage of this very simple solution: 50 or more percent more room for your data and --- in case of two disks --- in case both disks report different data (write error): which disk is correct? Some vendors name disk mirroring as disk duplexing of two or more controller are involved. Even in case a controller fails there is redundancy. To increase data security RAID 1 is often combined with some other RAID level. RAID 2 - Stripe Set Similar to RAID 0 all data is distributed to two or more disks. But an additional disk is used to write a hamming code. A hamming code detects errors and may repair smaller errors. Because a seperate disk is used to store the hamming code RAID 2 is slow. And because modern disks use a kind of error correction code by themself RAID 2 is now obsolete. RAID 3 - Stripe Set A performance-optimized aternative to RAID2. It's working on a stripe set and write errorcorrecting to a seperate disk, too. The difference is that a XOR-operations is used as the errorcorrecting code. When a disk fails it's contents may be calculated from the data of all other disks. There is no dataloss. RAID3 is fast but it's speed is reduced by transfering small, non-continuous blocks. Good for large transfer sizes. RAID 4 - Sector Striping RAID 4 does striping on basis of sector instead of bytes as RAID 3 does. There is also a seperate disk with checksum calculated by XOR-operations. Because this seperate disk is a bottle-neck RAID 4 isn't as fast as expected, at least when writing or in case of a disk failure (you don't need the XOR-values when reading). RAID 5 - Sector Stripng, Distributed Checksums This RAID level combines features from RAID 0, 3 and 4. Data is striped in sector-sized units on several (at least three) disks. Error correction is achieved by using XOR-operations. The calculated checksum isn't saved on a seperate disk but distributed along with the data sectors. Therefore there is no single disk acting as a bottle-neck. RAID 5 combines high security and good performance and is used for this reasons. Because the XOR-operation over large amounts of data is still something to take it's time RAID 5 is often implemented as hardware RAID with a special controller using a seperate unit to calculate it. Besides allowing to continue working with a failed disk RAID features some additional techniques to restructure redundancy after a failure: Hot Swapping Change a disk while keeping the machine up to replace a failed device. The failed device is stopped, changed and a replacement is automatically setup and filled to contain the data once stored on the failed device. There is no need to shutdown a running server. Be aware that hot swaaping need special connectors to avoid electrical damage. Hot Standby Add an additional disk to the system which is used to replace a failed device without sysop interaction. Until it is needed the disk is stopped. Reference Translated from german HDDFAQ (copright: PC POWER GmbH, Holger Ehlers@2:241/1050 aka 2:241/1020.20), available from 2:2474/424 as hddfaq.zip).