一种数据网格容灾存储模型及其数据失效模型

可靠性较高的数据网格多采用双副本容灾可以保证节点在灾难发生时进行有效恢复，但由于节点数据存储量较大，当节点发生灾难时，从一个节点进行数据恢复速度较慢，导致在数据恢复期备份节点发生失效的概率较大。针对这些问题，本文给出一个容灾存储模型，基于该模型推导出一个数据失效模型，理论证明该失效模型的数据失效概率明显小于双副本容灾方式的数据失效概率，同时在灾难发生时又能达到较快的恢复速度。设计了仿真实验，将模型的数据失效概率与双副本失效概率进行了比较，实验结果与理论推导一致，证明了存储模型和数据失效模型的正确性和有效性。最后给出进一步研究思路。

A Disaster-Tolerant Storage Model and a Low Data Failure Model for Data Grid

With double copies(DC) for Disaster-Toelerance,the data grid with higher reliability can ensure the effective recovery,reduce the data failure probability(DFP) when the disaster occurs,and can provide greater network bandwidth.But because of the large data storage of the nodes,when the disaster occurs to the nodes,data recovery from one node is a little slow,which leads to the higher failure probability of the backup nodes during the data recovery period,therefore,the data failure probability gets higher.To...