求解Hadoop作业调度问题的混合遗传算法  

A hybrid genetic algorithm to solve the problem of Hadoop job scheduling

作　　者：王丽红 夏魁良 金丹 WANG Li-hong;XIA Kui- liang;JIN Dan(School of Computer and Information Engineering,Heihe University,Heilongjiang Heihe 164399, China;Academy of Fine Arts,Heihe University,Heilongjiang Heihe 164399,China)

机构地区：[1]黑河学院计算机与信息工程学院,黑龙江黑河164399 [2]黑河学院美术学院,黑龙江黑河164399

出　　处：《齐齐哈尔大学学报（自然科学版）》2018年第3期6-10,共5页Journal of Qiqihar University(Natural Science Edition)

基　　金：黑龙江省教育厅科研业务费青年创新人才研究专项(2017-KYYWF-0360);黑河学院横向课题(HX201702)

摘　　要：将自适应遗传算法和改进的蚁群算法融合用以求解Hadoop作业调度问题。首先利用自适应遗传算法的全局搜素能力产生任务所分配的资源列表,在遗传算法的搜索速度逐渐降低时,适时切换到蚁群算法,由自遗传算法求解的最优解生成蚁群算法的初始信息素分布。改进蚁群算法的目标节点选择策略,考虑节点完成任务的成功率,加快蚁群算法求解最优解的速度。仿真结果表明,与遗传算法和蚁群算法相比,混合遗传算法用时较少,并且任务数越多,优势越明显。A hybrid optimization algorithm is proposed for Hadoop job scheduling problem, which is based on the combination of adaptive genetic algorithm and improved ant colony algorithm. The list of resources allocated by the task is generated by the global searching ability of the adaptive genetic algorithm.The genetic algorithm suspends and the ant colony algorithm starts at the optimal time.The algorithm gets the initial pheromone distribution using adaptive genetic algorithm. The selection strategy of target node is improved to accelerate the speed of ant colony algorithm to solve the optimal solution. Experimental results show the algorithm excels genetic algorithm and ant colony algorithm in performance,and it is discovered that the more number of the tasks, the better the algorithm performs.

关 键 词：遗传算法 蚁群算法 Hadoop作业调度 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]