MgdFlow:微电网场景下的多粒度数据流管理算法  

MgdFlow:multi-granularity data flow management algorithm in microgrid scenario

作　　者：荆有波 曹清越 朱瑞 JING Youbo;CAO Qingyue;ZHU Rui(Henan Institutes of Advanced Technology,Zhengzhou University,Zhengzhou 450003,China;Institute of Microelectronics,Chinese Academy of Sciences,Beijing 100029,China)

机构地区：[1]郑州大学河南先进技术研究院,河南郑州450003 [2]中国科学院微电子研究所,北京100029

出　　处：《通信学报》2023年第10期94-102,共9页Journal on Communications

基　　金：国家重点研发计划基金资助项目(No.2018YFC0823900)。

摘　　要：针对软件定义网络(SDN)设备被动响应式插入流表项导致表项空间占用和资源调度频繁的问题,提出了一种基于业务功能的多粒度数据流管理算法(MgdFlow)。通过延迟排序和多重聚合与分流,减少了流表项插入的数目和调度指令的次数。实验数据表明,所提算法在负载均衡性能方面比自适应两级流表提升18%,在平均表项占用方面比传统OpenFlow方案提升9%,在定义的控制器调度资源比率ImproveQoS方面比2种对比方案提升24%和12%。Aiming at the problem of frequent entry space occupation and resource scheduling caused by passive response insertion of flow entries by software defined network(SDN)devices,a operational function-based multi-granularity data flow management algorithm(MgdFlow)was proposed.Through delayed sorting and multiple aggregation and splitting,the number of flow entry insertions and scheduling instructions were reduced.Experimental data shows that the load balancing performance is 18%higher than the adaptive two-stage flow,and the average entry occupation is 9%lower than the traditional OpenFlow scheme.There are also 24%and 12%improvements in the defined controller scheduling resource ratio ImproveQoS compared to other schemes.

关 键 词：软件定义网络 流表管理 负载均衡 服务质量 

分 类 号：TP393.02[自动化与计算机技术—计算机应用技术]