A High Reliable and Performance Data Distribution Strategy: A RAID-5 Case Study

Abstract

With the development of the storage medium, such as the emerging of the SSD, the tradition way of data distribution can't keep up with the pace of the storage device development. Specifically, for example, Traditional RAID only enhanced the performance or reliability in the single storage medium and all the data are distributed in the same storage medium. But with the wide range of using raid, the problem of RAID data adventure has come to the surface gradually. Furthermore, under the mode of RAID-5 read modify-write (R-M-W), operation system collapse or the storage medium write error would lead to the damage of data, which could not be recovered and this problem is named write hole. According to the above problem, this paper proposes a new data distribution strategy, which is called high reliable and performance data distribution strategy (RPDD). RPDD use the high performance storage medium as a cache which is named storage cache. The most of the data is wrote to the storage cache and then distribute the data to the other logic relationship storage medium. As a result, the data is distributed in a hybrid storage medium. In order to take a better analysis of the RPDD in the next, we have combined RPDD in the RAID-5, which is named RAID-5-RPDD. RAID5-RPDD employs the dynamic scheduling mechanism, data transfer and hybrid consistency mechanism between the SSD and magnetic media storage devices to resolve the problem. In consequence, it can protect the RAID stripe data consistency under the condition of R-M-W and increase the I/O performance without the impact of the cost. In addition, it can improve the reliability of the RAID-5. The simulation test results show that RAID-5-RPDD I/O performance increase at the range of 9% with little penalty of the resource consumption.