RAID for Enterprise Computing

RAID has become commonplace in server computing environments. Today, most disk subsystems incorporate RAID technology to enhance their I/O performance and data availability. Software, or host-based RAID is also available from several vendors, including VERITAS. System administrators, application designers, and others responsible for implementing their organizations’ storage strategies are faced with a number of choices. This paper outlines the basics of RAID technology, describes the advantages of the various RAID alternatives, and lists other considerations in developing technical strategies for enterprise storage.


What’s in a Name?

RAID is an acronym for Redundant Array of Independent Disks:

• Redundant means that part of the disks’ storage capacity is used to store check data that can be used to recover user data if a disk containing it should fail.
• Array means that a collection of disks are managed by control software that presents their capacity to applications as a set of coordinated virtual disks. In host based arrays, the control software runs in a host computer. In controller based arrays, the control software runs in a disk controller.
• Independent means that the disks are perfectly normal disks that could function independently of each other.
• Disks means that the storage devices comprising the array are on-line storage. In particular, unlike most tapes, disk write operations specify precisely which blocks are to be written, so that a write operation can be repeated if it fails.