云计算下MapReduce多组容错机制架构的分析与研究  

Analysis and Study on under Cloud Computing Multiple Sets of Fault Tolerance Strategy Architecture of MapReduce

作　　者：张治斌[1] 李燕歌[1] 

机构地区：[1]河南理工大学计算机科学与技术学院,河南焦作454000

出　　处：《微电子学与计算机》2014年第1期52-55,59,共5页Microelectronics & Computer

基　　金：河南省软件科学研究(102400450064)

摘　　要：提出了MapReduce多组容错机制,在传统的Hadoop MapReduce架构上进行改进,即在同机柜中的TaskTracker节点之间增加了多组关系,这样可以缩短发现失效节点的时间,同时减轻JobTracker节点的负荷,减低了带宽使用率,减少网络拥塞.通过实验证明,MapReduce多组容错机制提高了MapReduce的工作效率.This paper presents a Multiple Sets of Fault Tolerance Strategy of MapReduce, improve Hadoop MapReduce framework, between TaskTracker node in the same cabinet group relationship added multiple sets relations, it can short the time of founding the failed node, at the same time, reduce the load of JobTracker node, bandwidth utilization and network congestion. The experimental results show that Multiple Sets of Fault Tolerance Strategy of MapReduce improved the efficiency of MapReduee.

关 键 词：云计算 MapReduce并行编程模型 多组容错机制 

分 类 号：TP302.8[自动化与计算机技术—计算机系统结构]