Comparison on Binary MDS Array Codes for Single Disk Failure Recovery
Junrong Ze, Hanxu Hou, Yunghsiang Sam Han · Proceedings of the 2018 2nd International Conference on Computer Science and Artificial Intelligence · 2018
Binary Maximum Distance Separable (MDS) array codes are widely used in storage systems such as RAID systems that can provide fault tolerance with minimum storage and low computational complexity. As data loss is normal in storage system, especially in large-scale distributed storage systems, recent studies have been focused on repair performance. This paper implements a class of binary MDS array codes of which the amount of network bandwidth required in repairing a failure is much less than that of the existing repair methods with three parity columns. We conduct experiments and present the performance comparison with other existing works. We show that the implemented binary MDS array code can save 40% repair bandwidth compared to Cauchy Reed-Solomon codes. Compared to the existing repair method X-code, the implemented binary MDS array code can save 20% repair time and 20% the amount of network bandwidth required in a repair process, respectively.