计算密集型与数据密集型混合网格作业调度算法  

Three grid job scheduling methods heuristics for data-intensive and computing-intensive jobs

作　　者：郝永生[1] 卢俊文 刘冠峰[3] 温娜[4] 

机构地区：[1]南京信息工程大学网络信息中心,江苏南京210044 [2]厦门理工大学,福建厦门361024 [3]苏州大学先进数据分析研究中心,江苏苏州215006 [4]南京信息工程大学大气科学学院,江苏南京210044

出　　处：《计算机工程与科学》2014年第8期1423-1429,共7页Computer Engineering & Science

基　　金：福建省教育厅科技项目B类(JB09199);国家自然科学基金资助项目(41005048);科技部资助项目(GYHY201106037;GYHY200906023)

摘　　要：针对计算密集型作业与数据密集型作业混合情况,在一个作业有时间限制的动态环境中,对传统的网格作业调度方法进行扩展,提出了三种网格作业调度启发式算法:Emin-min、Ebest、Esufferage。并在一个由多个Cluster组成的、通过高速网络连接的网格模型上,对三种算法进行验证。与Min-min算法的比较结果显示:三种算法均优于Min-min算法。与ASJS算法比较结果显示:Emin-min减少了等待时间与作业的makespan;Esufferage算法以减少作业完成量为代价,减少了作业的等待时间及makespan;Ebest在完成作业数量上与ASJS基本保持一致,但却增加了作业的等待时间与makespan。总体上,Emin-min具有比较大的优势。Most of the existing grid job scheduling methods focus on either data-intensive jobs or computing intensive jobs.In the dynamic environment where every job has its deadline,we extend the traditional grid job scheduling methods to propose three new grid job scheduling methods：Emin-min,Ebest and Esufferage.The three methods are validated on the Grid model with clusters that are connect ed by the high speed network.Simulation results demonstrate that our proposed methods are better than Min min.The comparison between the three methods and ASJS shows that,Emin-min reduces the wait ing time and makespan,Esufferage reduces the waiting time and the makespan greatly with the sacrifice of some jobs,and Ebest gives the same performance in unfinished jobs but has a larger value in waiting time and makespan than ASJS.In general,Eminmin has a better performance than Min-min and ASJS.

关 键 词：计算密集 数据密集 作业调度 平均执行时间 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]