基于可信任度的任务调度策略研究  

Research of Task Dispatching Strategy Based on Credibility

作　　者：邢艳芳[1] 秦军[1] XING Yanfang;QIN Jun(Nanguang College of Communication University of China,Nanjing 211172)

机构地区：[1]中国传媒大学南广学院,南京211172

出　　处：《计算机与数字工程》2020年第9期2190-2194,2259,共6页Computer & Digital Engineering

基　　金：江苏省高等学校自然科学研究面上项目(编号:17KJB510054,18KJB520053)资助。

摘　　要：随着云计算的快速发展,如何高效地进行云任务调度逐渐成为云计算研究的重点。云环境中由大量资源节点组成的集群具有异构性、动态性和不确定性等特征,节点出现故障的事件不可避免,因此需要将任务分配到可信任的资源节点上执行。针对MapReduce容错机制的缺点,论文提出引入节点失效恢复机制的可靠性任务调度策略,在进行任务分配时考虑节点失效的情况,避免任务被分配到运行期间会出现故障的节点上,造成资源浪费和增加了时间开销。对蚁群模拟退火算法进行扩展,将考虑失效恢复机制的可信任评估模型引入蚁群模拟退火算法中。通过仿真平台CloudSim,在不同任务数的情况下,比较FCFS算法和ACOSA算法任务执行成功率和目标函数值的大小,验证了任务调度算法和调度模型的有效性和稳定性。With the rapid development of cloud computing,to efficiently dispatch cloud tasks has gradually become the focus of cloud computing research.In a cloud environment,a cluster composed of a large number of resource nodes is characterized by heterogeneity,dynamics and uncertainty,and it is inevitable that nodes will fail,so tasks need to be assigned to the trusted resource nodes for execution.Aimed at the disadvantages of MapReduce’s fault-tolerant mechanism,this paper proposes reliable task dispatching strategy with the node failure recovery mechanism,considering node failure when dispatching tasks,to avoid tasks assigned to the failure node during the runtime,so as to cause the waste of resources and increase the time cost.The ant colony simulated annealing algorithm is extended and the trust evaluation model considering the failure recovery mechanism is introduced into the ant colony simulated annealing algorithm.Through the simulation platform CloudSim,the task execution success rate and the value of target function between FCFS algorithm and ACOSA algorithm are compared under different task numbers,and the effectiveness and stability of task dispatching algorithm and model are verified.

关 键 词：云计算 资源节点 任务调度 CloudSim ACOSA 

分 类 号：P315.69[天文地球—地震学]