RAID is an acronym for Redundant Arrays of Independent Disks. This storage technology has the potential to revolutionise the way on-line data is stored in everything from PC’s, to client/server, to mainframe computers. RAID technology provides superior storage performance, while offering improved availability, reliability and maintainability of information on disk.

A RAID array starts with at least two drives and continues up to very large arrays capable of providing massive amounts of disk storage.

RAID Advantages

The array can have a number of drives, but the host sees the array as one large disk through the RAID controller. In the event of a single drive failure, RAID’s redundancy keeps the data stream intact and the system up and running. Because the data is spread out over many independent disk, the effective seek time for finding records can be substantially reduced through multiple, simultaneous access of different records on different disks. RAID helps reduce or eliminate host resource allocation problems, because the data can be distributed across the disk array.

How Does It Work?

AMC provides hot-swap disk drive modules. These modules are a key component in the RAID strategy. Suppose you have a RAID system comprised of five modules in which one drive fails. The RAID system continues to provide the data that was on the failed drive through the redundancy on the other drives. This redundancy could be a copy of the data or data mathematically reconstructed from the four surviving drives. The system does not need to be powered down because the RAID enclosure supports hot-swapping of modules. The module containing the bad drive or power supply is removed and a new configured replacement module is inserted. The system returns to full redundancy, reincorporating the new drive by rebuilding the data on it without interrupting the data stream.

Different Approaches To RAID Technology

There are a number of different RAID approaches; these are the most popular methods.

Raid 0 Diagram

RAID 0. Disk stripping offers superior speed because it simultaneously access all drives through concurrent parallel read and writes; however, it does not offer redundancy. This RAID level is suitable, for example, for temporary synchronised video files.

Raid 1 Diagram

RAID 1. Disk mirroring. Two identical disks operate side-by-side in parallel, providing a complete, continuously updated backup of all data. Should one of the drives fail, the surviving mirrored drive provides read and writes. A replacement module (with new drive or power supply) can be hot-swapped. Data is backed-up to the new drive and full redundancy resumes. Disk mirroring offers the highest level of "mission critical" redundancy. This RAID level is suited for moderate-size file servers.

Raid 3 Diagram

RAID 3. Parallel Data Access offers the potential for extremely high data exchanges rates because rates because every disk in the array is used for each read and write. Only one I/O request can be active at a time but RAID 3 achieves exceptional speed because, rather than writing all the data to one disk, the individual record’s data is striped across multiple disks. For example, if you have a five-disk array, the data stream is logically broken up and continuously written or read across the four drives while the fifth drive provides redundancy by tracking parity. At each write, the parity is recalculated and rewritten on the fifth drive. The parity drive provides the missing numbers if one of the data drives fails; or, if the parity drive fails, the data drives can provide the data directly.

When a new module containing a configured drive or power supply is hot-swapped in, the array automatically rebuilds its redundancy. This RAID level might be used for larger file servers or applications requiring large sequential transfers.

Raid 5 Diagram

RAID 5. Independent Data Access is designed for high transaction throughput and concurrent access of data. RAID 3, with parallel data access, kept the data on four disks with the parity on the fifth. RAID 5 puts the entire record on one of the five disks and the related parity information on one of the four remaining disks. If any of the five disks fails, the parity information from the other four disks enables the system to mathematically rebuild the missing data. The data on the undamaged disk is simply read and written directly. This RAID level is ideal for database severs where there are greater amounts of independent access.