Abstract: This paper designs a database cluster architecture based on middleware technology to meet users’ high-input and high-output applications, and proposes a load balancing algorithm based on Markov model for the database cluster. The algorithm forecasts the load information of all execution nodes using Markov model within the period of sampling execution node load information, and balances the loads on all executor nodes using the above load prediction results. Experimental result shows that the proposed algorithm can improve the performance of database cluster effectively.

Key words: database cluster, load prediction, load balancing, Markov model, middleware, high-input and high-output

摘要： 为满足用户对数据库集群系统高输入高输出应用的需求，设计一种采用中间件技术的数据库集群系统，并针对该系统提出一种基于Markov模型的数据库集群负载均衡算法。该算法在执行节点负载信息采样周期内，利用Markov模型预测集群系统各执行节点的负载信息状态，根据预测的执行节点负载信息对集群系统进行负载均衡。实验结果表明，该算法能够有效提高数据库集群的性能。

关键词: 数据库集群, 负载预测, 负载均衡, Markov模型, 中间件, 高输入高输出

CLC Number: 

• TP311