复杂网络大规模数据流均衡调度方法  

Research on Balanced Scheduling Method of Large Scale Data Flow in Complex Network

作　　者：韩茂玲[1] HAN Maoling(Department of Information Engineering,Yantai Vocational College,Yantai 264670,China)

机构地区：[1]烟台职业学院信息工程系,山东烟台264670

出　　处：《成都工业学院学报》2021年第3期38-42,共5页Journal of Chengdu Technological University

摘　　要：目前对大规模数据流识别存在误检和漏检现象,导致大规模数据流在调度时的平均宽带利用率低,为此提出一种复杂网络大规模数据流均衡调度方法。采用SFlow技术,收集和检测大规模数据流信息,利用数据包的阈值判断数据流是否属于大规模数据流,并对数据包的阈值划分界限,降低识别时的误检和漏检概率;在识别的基础上,计算大规模数据流的传输顺序,确定其调度顺序,达到均衡调度的目的。然后在同一参数下,固定数据流携带任务数,通过改变数据流总条目和迭代次数,检测对比不同数据流条目下的平均带宽利用率和负载均衡度。实验结果显示,随着数据流总条目和迭代次数的变化,该方法的平均带宽利用率高,可以均衡虚拟机产生的负载。In the current data flow scheduling process,there are false and missing detection phenomena in the identification of large-scale data flow,which leads to the low average broadband utilization rate of large-scale data flow in scheduling.Therefore,a balanced scheduling method for large-scale data flow in complex networks was proposed in this paper.To identify large-scale data stream,sFlow technology was used to collect and detect the information of large-scale data stream.The threshold value of data packet was used to judge whether the data flow belongs to large-scale data flow,and the threshold value of data packet was divided to reduce the probability of false detection and missing detection in the identification of large-scale data flow.Based on the identification result of large-scale data flow,the transmission sequence of large-scale data stream was calculated,and the scheduling order was determined to achieve the goal of balanced scheduling.The experimental results show that,with the change of the total data flow entries and the number of iterations,the average bandwidth utilization of this method is high,and the load generated by the virtual machine can be balanced.

关 键 词：复杂网络 数据流 负载均衡 大规模 调度 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]