ICRS: An Optimized Algorithm to Improve Performance in Distributed Storage System

Abstract

Erasure codes, such as Reed-Solomn (RS) and CRS Codes, are being extensively deployed in distributed storage system since they offer significantly higher reliability than data replication methods at much lower storage overheads. But RS and CRS codes always impose a huge burden on system's performance while encoding and decoding when they provide significant savings in storage space. This paper puts forward an optimized algorithm named ICRS (Improved CRS) based on erasure coding technology, which is committed to improve the security and the utilization of storage space. By studying existing high reliability and space saving rate of coding technology, we imported coding mechanism into distributed storage systems. We have verified ICRS algorithm by theory analysis and simulation test. Through theory analysis, we can conclude that ICRS algorithm can improve the performance of encoding and decoding because they can shorten the computation times. We apply ICRS algorithm into our storage system model named Robot to test the performance. At the same time, we compare RS codes and CRS codes in Robot. The test results show that decoding speed can rise up nearly two times than the past serial decoding speed.