基于任务聚类的IABC云计算资源调度方法  

IABC clouding computing resource scheduling method based on task cluster

作　　者：郑洲[1] 张曦煌 张伟 ZHENG Zhou;ZHANG Xi-huang;ZHANG Wei(Information Center, Wuxi Institute of Arts and Technology, Wuxi 214206, China;Internet of Things College, Jiangnan University, Wuxi 214122, China)

机构地区：[1]无锡工艺职业技术学院信息中心,江苏无锡214206 [2]江南大学物联网学院,江苏无锡214122

出　　处：《计算机工程与设计》2018年第12期3755-3761,3822,共8页Computer Engineering and Design

基　　金：国家自然科学基金项目(61302133)

摘　　要：为提高云资源调度的性能,提出基于任务聚类的改进人工蜂群调度方法。建立云任务模型、云资源模型、约束模型、服务质量模型等云问题模型;改进跟随蜂蜜源选择概率,提高优质解选择概率和其附近的有效搜索,改进雇佣蜂的蜜源搜索策略,使算法能够兼顾大范围搜索并快速收敛;提出任务聚类和资源评价方法,将云任务和云资源都分为计算型、通信型、存储型3类,有针对性地进行资源分配。经仿真验证,通过调整优化系数改变优化重心,改进人工蜂群算法在花费、任务完成时间、可靠性方面均可取得最优解;基于聚类的调度方法在任务完成时间和资源利用率上具有优势,任务量越大优势越明显。To improve property of cloud resource scheduling,the scheduling method based on improved artificial bee colony algorithm and task cluster was proposed.Cloud task model,resource model,restriction model,and service quantity model were built.Nectar source selective probability of following bee and selection probability of optimal solution were improved.Nectar source searching strategy of employing bee was improved,so that extensive search and rapid convergence were balanced.Task cluster and resource evaluation method were put forward.Cloud task and resource were divided to computational type,communication type and storage type,so that cloud resources were assigned to task pertinence.By simulation,optimization center can be changed by changing coefficient,and improved artificial bee colony algorithm possesses best cost,completion time and reliability.Completion time and resource utilization rate by task cluster are optimal,and advantage is more obvious with workload increasing.

关 键 词：云计算 资源调度 改进人工蜂群算法 任务聚类 虚拟资源评价 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]