An Efficient Distributed B-tree Index Method in Cloud Computing

To support online index and range queries, the Distributed B-tree is adopted to index the mass and rapidly increasing data in cloud computing. But current Distributed B-tree has three defects: low degree of concurrency, frequent node splitting and high cost of updates in clients. For above mentioned defects, this paper presents efficient distribute Btree index in cloud computing environment, which effectively enhances the performance of the distributed B-tree index. First, it improves concurrent access by the distributed B-tree high concurrency access method based on node split history. Second, it reduces the splitting frequency by the method of dynamic changing node size. Finally, it reduces node update cost in all client buffers by the regional delayed update method. Experimental results show that, this method has high performance in cloud computing environments.