基于欺负算法的改进选举算法

Improved election algorithm based on bully algorithm

欺负算法产生大量通信信息,时间开销大,占用系统资源过高,严重影响了分布式OLAP系统的性能。针对该问题,提出一种基于欺负算法的改进算法。该算法采用一对一的方式直接向性能最优的节点发送选举消息,以降低选举过程中产生信息的通信量和选举时间开销;并通过循环选举保证选举出系统的最优节点担任系统协调者。实验结果表明,该改进算法有效地降低了消息通信量,减少了时间开销,能更好的应用于分布式OLAP系统。

Bully election algorithm has the high traffic complexity, generates a large number of messages, and takes up system resources. It affects the performance of distributed OLAP system. To address the problem, an optimized election algorithm based on the bully algorithm is proposed. The algorithm directly sends election messages to nodes with the optimal performance by one-to-one. And the optimal node is elected to act as the system coordinator by circular election. The tested results show that the improved election algorithm can reduce the messages and the time overhead effectively, and can be applied in distributed OLAP system well.