Computer Science and Engineering, Department of
Date of this Version
2012
Citation
Department of Computer Science & Engineering, University of Nebraska-Lincoln, Technical Report, TR-UNL-CSE-2012-0004
Abstract
MapReduce has been widely used as a Big Data processing platform. As it gets popular, its scheduling becomes increasingly important. In particular, since many MapReduce applications require real-time data processing, scheduling real-time applications in MapReduce environments has become a significant problem. In this paper, we create a novel real-time scheduler for MapReduce, which overcomes the deficiencies of an existing scheduler. It avoids accepting jobs that will lead to deadline misses and improves the cluster utilization. We implement our scheduler in Hadoop system and experimental results show that our scheduler provides deadline guarantees for accepted jobs and achieves good cluster utilization.