The data reliability techniques in BC-KVDB

Key-Value DB has been become a novel distributed storage framework for cloud storage. It mainly focuses on solving the problems of horizontal scalability and high throughput for cloud storage. However, data reliability has become bottleneck problem for the distributed Key-Value DB. In this paper, we introduce a novel distributed storage system: BC-kvDB (Big Cloud key-value DB). In BC-kvDB, we propose multi-master mechanism to solve the problem of single point of failure for secondary master. Besides, we also introduce a novel flexible replica management technique for key-value data chunk. The experiment results expose that BC-kvDB performs better than Hbase on random read/write test. It also achieved a factor of 2.5 improvements in sequence write performance than Hbase.