基于HBase的大数据平台负载均衡算法分析与优化

摘要

In order to improve the performance of HBase database, the original load balancing algorithm of HBase is analyzed.The original algorithm strategy ensures that the number of Region per Region Server is the same, but because the access of each Region is not equal, it may cause the load imbalance, generate hot data, and affect the response efficiency of the request.Part of the Region will become a hot spot, which some Region Server overload.It is very important to consider the impact of the heat of the data on the load balancing of a distributed database.Therefore, the prediction method is used to design the HBase load balancing algorithm and predict the heat of the data as the load of the Region Server.The experiment shows that the optimized load balancing algorithm can improve the performance of HBase database.%HBase原有负载均衡算法策略是保证每个Region Server的Region数目相同.但每个Region的访问并不均等, 可能造成负载不均衡情况, 产生热点数据, 影响请求响应效率.部分Region会成为热点, 导致一些Re-gion Server负载过重.为提高HBase数据库整体性能, 将数据热度因素考虑在内, 使用预测方法设计HBase负载均衡算法, 把预测数据的热度作为Region Server负载.实验表明, 优化后的负载均衡算法可在一定程度上提高HBase数据库性能.