International Journal for Advance Research in Engineering and Technology (IJARET)
In this paper, the authors present the study of MapReduce performance in Hadoop architecture, the job assigned to the Hadoop is divided in to tasks among the node of the cluster. But some of the nodes may be running slowly in the cluster to result in slow processing of the MapReduce because of their processing capability of load on the node. Such tasks that are assigned to the slow running node in the cluster are efficiently backed up on some other nodes. There are some existing strategies to backup the slow running tasks on the alternate machines. This paper emphasizes considering data locality and data skew.