Cloud computing is a power platform to deal with big data. Among several software frameworks used for the construction of cloud computing systems, Apache Hadoop, which is an open-source software, becomes a popular one. Hadoop supports for distributed data storage and the process of large data sets on computer clusters based on a MapReduce parallel processing framework. The performance of Hadoop in parallel data processing is depended on the efficiency of a job scheduling algorithm underworking. In this paper, we improve the performance of the well-known fair scheduling algorithm adopted in Hadoop by introducing several mechanisms. The modified scheduling algorithm can dynamically adjust resource allocation to user jobs and the precedence of user jobs to be executed. Our approach can properly adapt to the runtime environment’s condition with the objective of achieving job fairness and reducing job turnaround time. Performance evaluations verify the superiority of the proposed scheduler over the original fair scheduler. The average turnaround time of jobs can be largely reduced in our experiments.
展开▼
