I/O Performance Cost Estimation Technique for SSD based Cloud RAID

초록

The purpose of this paper is to narrow down difference between ideal and real IO performance of all possible configuration for cloud storage system which relies on input parameters such as chunk size. We propose new configuration architecture and cost estimation technique for enhancing IO performance cost of SSD based cloud storage system(SCSS). The function of parameter manager evaluates cost value using SAD function between maximum and selective throughput in terms of various chunk sizes, RAID and file system types. The experimental results show that among all possible configurations maximum read performance achieved is 2.75 GB/Sec for RAID-5 and maximum write performance is 1.26 GB/Sec for RAID-6 using proposed method in 4-bay