| Processing power has increased dramatically through Very Large Scale Integration technology and parallel architectures in computer system. As processing power increases, so does the demand for increased Input/Output performance. But the access speed to hard disk which is the major storage device in computer system has increased very slowly. Two decades ago, the conception of Redundant Arrays of Inexpensive Disks (RAID) was introduced by David Patterson in University of California at Berkeley. In RAID system, parallel is used to raise data-access rate and redundancy is added in to improve system's reliability.Although today single disks are highly reliable, when a disk array consists of hundreds of even one thousand disks, the probability that at least one disk will fail within a day or a week is high. How to protect data against catastrophic disk failure has become a crucial issue in the design of very large disk arrays. With the increase of the disk array scale, tolerating more disk failures becomes more and more important. It will increase the reliability of storage system when coding techniques are applied in system, especially in the RAID system.The main contents and results are as follows.1. The features and main implementation methods of RAID are summarized. The coding methods in traditional RAID are studied and the new development of erasure-correcting codes which can be applied in RAID is generalized.2. Another proof of the Maximum Distance Separable property of X codes is presented. Based on this proof, the decoding algorithm is described in this paper.3. Four typical coding shemas in RAID, EVENODD, X codes, STAR and RS, are simulated in software; their feasibility and correction are demonstrated in the simulation. Their performances are analyzed respectively and their advantages and disadvantages are also discussed. |
PDF Full Text Request