基于Hadoop集群的多表并行关联算法及应用  

Multi-chart parallel correlation algorithm and application based on the Hadoop cluster

作　　者：郑晓薇[1] 马琳[1] 

机构地区：[1]辽宁师范大学计算机与信息技术学院,辽宁大连116081

出　　处：《微型机与应用》2013年第4期91-93,共3页Microcomputer & Its Applications

摘　　要：针对因特网环境下并行数据库实现多个大数据表关联存在的计算瓶颈,基于Hadoop集群设计了一个并行关联多个大数据表的简便算法MR_Join。以商业网站凡客诚品的销售数据为例进行实验,验证算法的可行性并做出应用实例。实验结果表明,MR_Join算法可以有效地实现大数据表的快速关联,具有显著的并行效率。In the environment of Internet, using parallel database to realize the data table connection has calculation bottlenecks. Based on the Hadoop cluster, design a parallel correlation muhiple large data table MR_Join simple algorithm. In order to further verified the feasibility of the algorithm, in the Hadoop cluster to commercial websites where sales data as an example for the experiment. The experimental results show that MR_Join algorithm can effectively achieve the large data table fast connection, has the remarkable parallel efficiency.

关 键 词：HADOOP集群 Mapreduce编程模式 MR_Join算法 数据表并行关联 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]