RAID (Redundant Array of Independent Disks) is a complex and sophisticated disk drive subsystem that uses two or more hard disks and a RAID disk controller which act together as a single storage system. They are commonly implemented by large organisations in a server based environment with the aim to manage and protect large volumes of mission critical data, increase performance while providing a high level of fault tolerance.
RAID can improve performance due to "disk striping", a process which allows multiple disks to read and write simultaneously. RAID systems can provide fault tolerance by processes called "data mirroring" and "data parity" – whereby data is entirely copied onto two drives and can be used to rebuild lost data when a drive fails.
While there are many types of configurations, sizes and platforms, most RAID systems can be assigned one of the following groupings (RAID 0, 1 and 5 being the most common levels):
|
RAID Level 0 |
Disk Striping Without Parity
Disk devices are organised and written equally and alternatively (striping) between the disks. There is no redundancy provided, but a higher level of efficiency and performance is achieved by evenly spreading the data across several drives. |
|
RAID Level 1 |
Disk Mirroring
Data is written to two disks simultaneously, enabling the user to achieve a high level of redundancy and ensuring complete data duplication on both disks and hence no data loss in the event of one drive failing. While this is a good entry level system, the system does not provide an advantageous storage capacity configuration as the storage cost is twice that of a single drive being used. (Minimum of 2 disk drives required) |
|
RAID Level 3 |
Parallel Transfer with Parity
Data is striped across several disk drives at byte level, with data parity being stored on one drive. (Requires a minimum of 3 disk drives). This allows data recovery in the event of failure of one of the disks, but also provides a high data transfer rate. |
|
RAID Level 4 |
Independent Data disks with shared parity disk
The data is striped amongst the disk drives, like RAID 0. In addition, redundancy data (parity information) is are stored on a separate disk drive like RAID 3. |
|
RAID Level 5 |
Independent Data disks with distributed parity blocks
The data is striped at block level across several drives, with redundancy information (parity data) being distributed amongst the drives. There is no separate parity drive, and hence no write bottleneck often caused in RAID 4 systems. |
As the number of disks that are used increase, so do the number of combinations that can be created using the basic five RAID levels. For example For example RAID 6, this is RAID 5 extended with a parity block utilizing block level striping with two parity blocks being distributed across all of the disks available in the subsystem. The core feature of RAID 6 is the ability is to provide a further level of protection against data loss when rebuilding an array. They can incorporate different segments, levels and arrays using different configurations of disk mirroring, striping and parity.