RAID is an acronym for Redundant Array of Inexpensive Discs and was originally conceived utilising SCSI Drive controllers. Initially SCSI interfaces were the only controllers that had the ability to control more than two drives but now multi-channel IDE and Serial ATA (SATA) controllers are quite a common and, at times, an inexpensive alternative to SCSI.
There are quite a few types of RAID arrays, one of which is a non-redundant array that offers greater performance but no data protection.
RAID-0. This array type utilises striping across two drives for example to provide better performance than two individual drives. There is no redundancy of data and therefore no fault tolerance.
RAID-1. Commonly known as disk mirroring, RAID-1 consists of at least two drives each pair of which contains an exact duplicate of each other. There is no striping but should one of the drives fail the second drive in the pair holds a full copy of the data. Read performance is actually improved as either of the drive pairs can be read at the same time although write performance is the same as a single drive as pairs must be written simultaneously.
RAID-3. In this situation data is stripped and one of the drives in the array is dedicated to storing parity information. I/O cannot be overlapped as any I/O operation addresses all drives at the same time as a result RAID-3 is relatively slow.
RAID-4. The stripes used in RAID-4 implementations are relatively large so records can be read from a single drive and drive read operations can be overlapped to improve speed. However there is no overlapping of write operations as the parity drive must be updated with every write operation.
RAID-5. An improvement on RAID-4 as it includes a rotating parity array which means that all read and write I/O can be overlapped. RAID-5 does not store redundant data but by using the parity array information data can be reconstructed when a faulty drive is replaced. RAID-5 requires a minimum of three drives but typically five drives in an array.
RAID-6. Is actually Dual Parity and takes the parity region from RAID-5 and duplicates it so each disk has two parity regions that are calculated separately. As a consequence a RAID-6 array can recover from the loss of two drives, but performance is impacted and an extra drive is required which increases the cost of implementation.
RAID-ADG. Or Advanced Data Guard is HP's proprietary implementation and is similar to RAID-6 in that RAID-ADG has Dual stipes on each drive.
RAID-7. Technically not an official RAID level RAID-7 is a proprietary system that includes an internal CPU, real-time embedded operating system and caching in the array hardware. The approach improves upon RAID-4 by managing the array asynchronously and thus improving random read and write performance over RAID-4. A large cache is critical to the systems performance and must be supported by UPS to prevent data loss should a power failure occur.
RAID-10. This type offers a combination of RAID-0 (striped) and RAID-1 (mirroring). As a consequence the array has high fault tolerance of mirroring and the speed of a RAID-0 striped array. But the implementation can be expensive as twice as many disks are required as a standard RAID-5 implementation also scalability is limited.
RAID-53. This type combines striping by byte and parity across two disk sets and is fault tolerant and fast, particularly with large files. It is expensive to implement and like RAID-3 has poor utilisation of disk capacity.
