Difference between revisions of "RAID"
(Quick overview) |
(ECC location; example of striping in early UNIX) |
||
Line 5: | Line 5: | ||
==Types== | ==Types== | ||
− | Different RAID schemes are referred to by labels of the form 'RAID n', where n is a digit; all produce a single virtual drive. RAID 2-6 all use an [[error-correcting code|ECC]], and are able to recover from damage to a single physical drive | + | Different RAID schemes are referred to by labels of the form 'RAID n', where n is a digit; all produce a single virtual drive. RAID 2-6 all use an [[error-correcting code|ECC]], and are able to recover from damage to a single physical drive. In RAID 2-4, the ECC information is stored on a single other drive; in RAID 5 and 6, it is stored om multiple drives. |
The main ones are: | The main ones are: | ||
Line 23: | Line 23: | ||
* [https://www2.eecs.berkeley.edu/Pubs/TechRpts/1987/CSD-87-391.pdf A Case for Redundant Arrays of Inexpensive Disks (RAID)] | * [https://www2.eecs.berkeley.edu/Pubs/TechRpts/1987/CSD-87-391.pdf A Case for Redundant Arrays of Inexpensive Disks (RAID)] | ||
* [http://web.eecs.umich.edu/~michjc/eecs584/Papers/katz-2010.pdf RAID: A Personal Recollection of How Storage Became a System] | * [http://web.eecs.umich.edu/~michjc/eecs584/Papers/katz-2010.pdf RAID: A Personal Recollection of How Storage Became a System] | ||
+ | * [https://minnie.tuhs.org/cgi-bin/utree.pl?file=V6/usr/man/man4/rk.4 rk.4] - 'striping' in UNIX V6 | ||
[[Category: Device Basics]] | [[Category: Device Basics]] |
Revision as of 17:30, 20 August 2023
RAID (originally an acronym for 'redundant array of inexpensive disks'; later redefined to stand for 'redundant array of independent disks') is the name for a family of schemes to increase the reliability, performance or size of disk mass storage by using a number of drives in parallel.
The basic ideas pre-date the original definition of the term (below): early UNIX disk device drivers had split a single virtual drive across multiple physical drivers ('striping'); and mirrored disks (for reliability) had been used by Tandem.
Types
Different RAID schemes are referred to by labels of the form 'RAID n', where n is a digit; all produce a single virtual drive. RAID 2-6 all use an ECC, and are able to recover from damage to a single physical drive. In RAID 2-4, the ECC information is stored on a single other drive; in RAID 5 and 6, it is stored om multiple drives.
The main ones are:
- RAID 0 - a virtual drive is split across multiple physical drives; provides increased performance only
- RAID 1 - a virtual drive is mirrored across multiple physical drives; provides increased reliability only
- RAID 2 - a virtual drive is striped across multiple physical drives, a bit at a time
- RAID 3 - a virtual drive is striped across multiple physical drives, a byte at a time
- RAID 4 - a virtual drive is striped across multiple physical drives, a block at a time
- RAID 5 - a virtual drive is striped across multiple physical drives, with distributed ECC
- RAID 6 - a virtual drive is striped across multiple physical drives, with replicated, distributed ECC