Improving RAID storage system dependability and performance with HRDCC
Storage system is set up by using the basic capability of a storage device, such as adding layers of hardware and software to obtain a highly reliable, higher performance, and easily managed system. We present the design, implementation, and evaluation of the HRDCC (HUST-RAID Dual Controller based on Controller). To address the problem of single point of failure with a single RAID controller, HRDCC was devised to avoid data loss or even data corruption. Evaluations of the storage system dependability have focused mainly on disk arrays, including the dependability of array hardware and RAID mechanisms and the recovery time after failure. We use modeling technology, including failure tree theory and stochastic Petri nets, to evaluate the dependability and the performance of dual controller system. MTTF (Mean Time to Failure) and MTBF (Mean Time Between Failure) scale reliability of storage system. Utilization rate reflects speed matching of controller and storage device. The results show that, in addition to improving the reliability, dual controller design can improve the average utilization rate of string controller by selecting the channel transfer rate and the communication means between the array controller and the string controller properly, and indicate that the reliability and the performance of dual controller is higher than single controller.
Author's Name: Wan, Y., Feng, D., Wang, F., Liu, L., Yang, T.