A Task Scheduling Policy for Heterogeneous MapReduce Cluster
Chiu Chui-Ming, Huang Sheng-Wei, Huang Tzu-Chi, Shieh Ce-Kuen, Tsai Ming-Fong, Chen Lien-Wu · Frontiers in artificial intelligence and applications · 2015
MapReduce is a programming model and its associated run-time system proposed in 2004, which can process large scale of data in clusters with simple program logic. MapReduce has a potential problem running on a cooperative cluster which is combined with machines having different configurations. The problem will cause unexpected performance degradations and should be avoided. In this paper, a task scheduling policy is proposed to take higher utilization of all computing nodes in heterogeneous clusters.