TABLE OF CONTENTS The development of high-performance reliable computing has been the philosopher s stone of computing systems since the development of the first computers. The first pro-duction-ready electronic computer, ENIAC, used vacuum tubes for electronic switching and had several failures per day. Most of us have experienced the frustration of having our computer fail (usually just before giving a presentation, or halfway through writing an important document!). As long as computer systems have moving parts they will develop failures. Even machines without moving parts are subject to breakage. Even so, reducing the likelihood of an outage from a device failure remains a high priority for the industry and the introduction of RAID storage devices represented a massive leap forward in the reliability of storage systems. RAID stands for reliable array of inexpensive disks, though the inexpensive adjective is highly debatable. RAID devices are designed to offer performance and reliability through a combination of parallelism striping, redundancy, and parity.
The theoretical beginnings of RAID began at IBM in the late 1970s, when Normal Ouchi filed a patent for the system for recovering data stored in failed memory unit. The basic concepts of Ouchi s patent describe the most important ideas of RAID 5. RAID was later formalized for storage systems in the 1988 SIGMOD paper by Patter-son et al. [1988]. Patterson et al. s paper was considered the seminal paper introducing the RAID concept, and the disk array industry expanded rapidly after its...
