一种MapReduce系统能耗优化的任务调度算法  

The Energy Consumption Optimization Scheduling Algorithm for MapReduce System

作　　者：王珏[1] WANG Jue(Software School,East China Jiaotong University,Nanchang Jiangxi 330013,China)

机构地区：[1]华东交通大学软件学院,江西南昌330013

出　　处：《江西师范大学学报（自然科学版）》2024年第4期364-370,共7页Journal of Jiangxi Normal University(Natural Science Edition)

基　　金：江西省教育厅科学技术研究(180319)资助项目.

摘　　要：该文以能耗为目标,将一个作业中的任务调度问题表述为整数线性规划,并在遵循服务级别协议的前提下,通过优化Map和Reduce任务在slot上的调度策略来最小化系统的能耗,提出了一种能耗敏感型的任务调度算法.实验结果表明:与其他常用的算法相比,该算法在TeraSort、PageRank和k-means聚类工作负载上能够接近能耗的最小值.Taking energy consumption as the objective,the task scheduling problem in a job is formulated as an Integer Linear Programming and an energy-sensitive task scheduling algorithm is proposed by optimizing the scheduling strategy of Map and Reduce tasks on slots under the premise of adhering to the service-level agreement(SLA)to minimize the energy consumption of the system.Experimental results show that compared with other commonly used algorithms,this algorithm can approach the minimum value of energy consumption on TeraSort,PageRank,and k-means clustering workloads.

关 键 词：MAPREDUCE 大数据 整数线性规划 调度策略 

分 类 号：TP333[自动化与计算机技术—计算机系统结构]