摘要
提高云计算环境中计算节点的效率具有重要意义。文中分析了云计算并行编程模式MapReduce及其存在的不足,提出了结合DAG图(Directed Acycline Graph,DAG)进行任务分配的方法,对相同出口的路径上的子任务进行调整,控制Map阶段的中间处理结果提交时间,使Reduce阶段任务的推进不受影响。文中方法有利于提高节点的利用率,保持负载均衡,进行了仿真实验,结果也验证了算法的正确性。
It is important to improve the efficiency of nodes in cloud computing environment. In this paper,the model of MapReduce and its defects are analyzed. A method of task assignment based on DAG graph（ Directed Acycline Graph,DAG） is proposed. The sub-tasks on the same export path are adjusted,which make the intermediate processing results in the Map stage to be submitted for the same time. The task in Reduce phase is not affected. This method is advantageous to improve the utilization ratio of the node and keep the load balance. The experiments are carried out. The correctness of the algorithm is verified by the results.
出处
《信息技术》
2017年第9期76-78,83,共4页
Information Technology
基金
四川省教育厅资助项目(16ZB0412)
成都理工大学工程技术学院基金资助项目(C122015007)
