基于水平纠删码的云存储数据布局方法

针对云存储系统的扩展性和数据容错问题,设计了一种基于水平阵列纠删码的通用云存储系统方案,系统可按存储节点子集进行规模扩展；给出了一种可容3列随机删除错的数据布局方法,并利用校验数据位与信息数据位之间的对应关系,寻找到一种具有低计算复杂度的数据重构算法,提高了丢失数据的恢复性能.理论分析和原型系统测试表明,本布局方法与存储方案适合于构建对数据编译码复杂度、系统扩展性都具有较高要求的云存储系统.

A Data Placement Based on Level Array Codes in Cloud Storage

In order to content with the expansibility and fault tolerance of the cloud storage system, expandable encoded cluster storage architecture with low price and low data redundancy was presented in this article. The program associated with the file block encoding technology was introduced to build the high-availability clustered storage system which was able to tolerate triple simultaneous storage node failures in any erasure pattern. The system had the optimal storage efficiency and the remarkably fault-tolerant capability. At last, the theoretical analysis and the test of the cloud storage system was introduced in detail, and an intensive performance evaluation study on it was performed.