Replication-based data distribution method with RAID ( redundant array of independent disks) architecture causes the problems of load increase, reliability and accessing bandwidth reduction, etc.For these problems, we propose a parity check-based stripe mirror data distribution solution, referred to as RAID-P.First, the scheme copies the stripe data block from other disks and then associates every stripe with a parity check block, and accesses them by directing the read requests to appropriate mirror backup data, therefore improves the reliability of the array and access bandwidth, moreover, it also possesses the fault units positioning function which is necessary for RAID disks array.After comparing with other types in same structure by simulation, the results show that the model demonstrates a higher reliability and can provide significantly better I/O performance between the disk requests through the methods of data striping, block mirror and parity check blocks.%针对基于复制的RAID(独立磁盘冗余阵列)架构的数据分布方法所引起的负载增加、可靠性和接入带宽减少等问题,提出一种基于奇偶校验的条带镜像数据分布方案,简称RAID-P。首先,该方案在其他磁盘中复制条带数据块,然后将每个条带与一个奇偶校验块相关联,并通过将读取请求引导到合适的镜像备份数据来进行访问,从而提高阵列的可靠性和接入带宽,并且具备RAID磁盘阵列所必须的故障单元定位功能。与其他类此结构进行仿真比较后,结果表明:通过数据条带化、块镜像和奇偶校验块方法,该方案表现出了更高的可靠性,并能够在磁盘请求之间提供明显更好的I/O性能。
展开▼