RAID - Redundant Array of Inexpensive (or Independent) Disks - is an evolving storage technology that offers significant advantages in performance, capacity, reliability, and scalability to firms that have requirements beyond those offered by a single PC disk. A RAID system comprises three main components:
  • A RAID controller
  • An array of five or more disks
  • An enclosure that contains the controller, disks, and supporting power and cooling facilities.

The RAID controller is an electronic device that provides the interface between the host computer and the array of disks. From the viewpoint of the host computer, the RAID controller makes the array of disks look like one virtual disk that is very fast, very large, and very reliable. A SCSI RAID controller that is external to the host computer provides disk operation that is seamless and transparent to the host computer; i.e., the host computer does not require changes to software to realize the benefits a RAID system offers.

The performance of electronic devices such as microprocessors has grown at a rapid pace, yet the electromechanical design of computer disks has limited their performance growth. Indeed, microprocessor performance has been doubling about every two years, while disk performance has taken ten years to double. A RAID controller overcomes this limitation by using parallel data paths to read and write information to the disks in a RAID array. Thus, it performs the operations of reading and writing information to several disks simultaneously. With four data disks, for example, a RAID system can read and write information at a rate almost four times the rate of a single disk. Such high performance enables demanding applications, such as real-time video editing, to be accomplished that would otherwise be impossible or extraordinarily expensive. RAID controllers can organize data on the disks in several ways to offer performance advantages for different types of applications; the chart below classifies the most common methods.

RAID systems provide large amounts of storage by making the data on several disks readily available to the host computer. Sophisticated RAID systems provide scalability by allowing customers to daisy chain multiple disks on each data path emanating from the controller. A high performance RAID controller can address 90 disks in a fully configured RAID system.

RAID systems also provide high reliability and data availability through a technique called parity checking. In this scheme, when the RAID controller writes information onto the disks, it also writes redundant information called parity bits. Should a disk fail, this parity information enables the RAID controller to recompute the lost information as it is requested without degrading performance. Advanced RAID controllers will reconstruct the lost data onto a spare disk, so that the system can survive another disk failure.

With increasing demands for mass storage performance, capacity, and reliability, many firms are adopting RAID technology to complement their computer systems to support demanding applications such as digital imaging, prepress, on-line transaction processing, data warehousing, and file servers.

RAID Level Description Data Reliability Data Transfer Rate I/O Request Rate Application Strength Cost
RAID 1 All data copied onto 2 separate disks Very high. Can withstand selective multiple disk failures Data transfer rate is higher than single disk for reads, but does not offer load balancing Twice that of a single disk for reads. Slightly slower than single disk for writes General Very high. Requires twice as many disks for redundancy
RAID 2 Data striped across multiple disks with parity on multiple disks Very high. Can withstand selective multiple disk failures High if error correcting codes are computed by hardware Similar to twice that of a single disk General High. Requires multiple disks for redundancy
RAID 3 Data striped across all data disks with dedicated parity disk Much higher than single disk. Can withstand single disk failure Highest of all types listed here for reading and writing Faster than a single disk, owing to parallel disk accesses Video, prepress, medical imaging, and other large file applications Low. Requires only one disk for redundancy
RAID 4 Data striped across some data disks with dedicated parity disk Much higher than single disk. Can withstand single disk failure High compared to single disk for reads but significantly lower than single disk for writes* High compared to single disk for reads but significantly lower than single disk for writes* Predominantly read-oriented with few writes Low. Requires only one disk for redundancy
RAID 5 Data and parity striped across multiple disks Much higher than single disk. Can withstand single disk failure High compared to single disk for reads but lower than single disk for writes* High compared to single disk for read but generally lower than single disk for writes* Transaction processing with high read to write ratio Low. Requires only one disk for redundancy

* Write operations are slow in these cases because the controller must read parity information from a disk and recompute parity information for the disk before it writes information to the disk array.


This is a copy of an article published @ http://www.digidata.com/raiddesc.html/