Performance Analysis of Utilizing Reed Solomon Code in Redundant Array of Independent Disk

Abstract

The Redundant array of independent disks (RAID) is a method of storing the same data on multiple hard disks in different places. By placing data on multiple hard disks, I/O operations can overlap in a balanced way, improving performance. RAID utilizes multiple disks to achieve higher data reliability at the cost of reduced storage capacity. RAID6 has become a popular option for RAID disk array due to its capability to recover two disk failures in a setup with two parity disks. A RAID6 setup usually includes several data disks and two parity disks. Although the general setup is similar, in practice there are a variety of encoding scheme to achieve similar effect. Two commonly used coding schemes are Reed-Solomon (RS) code and EVENODD code. This article will discuss the effectiveness of RS code in theory, by calculating the data I/O operation needed to read, write and rebuild disk in both coding schemes, and compare the result against that of EVENODD coding. Simulation of RAID6 structure will also be performed, both under ideal condition and non-ideal condition, to consider the influence of real factor such as degrading disk has on the effectiveness of such RAID structures.