计算机工程与应用 ›› 2019, Vol. 55 ›› Issue (21): 115-121.DOI: 10.3778/j.issn.1002-8331.1808-0436

• 网络、通信与安全 • 上一篇    下一篇

面向NUMA架构的虚拟网络功能部署技术研究

许道强,邹云峰,邓君华,李浩,钱柱中   

  1. 1.国网江苏省电力有限公司,南京 210024
    2.国网江苏省电力有限公司 电力科学研究院,南京 210024
    3.南京大学 计算机科学与技术系,南京 210023
  • 出版日期:2019-11-01 发布日期:2019-10-30

Research on Virtual Network Function Chain Deployment for NUMA Systems

XU Daoqiang, ZOU Yunfeng, DENG Junhua, LI Hao, QIAN Zhuzhong   

  1. 1.State Grid Jiangsu Electric Power Co., Ltd., Nanjing 210024, China
    2.Electric Power Research Institute, State Grid Jiangsu Electric Power Co., Ltd., Nanjing 210024, China
    3.Department of Computer Science and Technology, Nanjing University, Nanjing 210023, China
  • Online:2019-11-01 Published:2019-10-30

摘要: 针对非一致内存访问架构的集群系统,探讨了虚拟网络功能链的时延问题,以及重用已部署的虚拟网络功能提升资源利用率机制。形式化定义了面向最小化时延的虚拟网络功能部署问题,证明其为NP难问题,提出了基于马尔可夫链概率迭代的部署算法MarVNFP。实验证明,与典型的虚拟网络功能部署算法MRP相比,MarVNFP算法能减少虚拟网络功能链近40%的跨域数据传输,有效降低了时延。

关键词: 网络功能虚拟化, NUMA, 资源调度, 马尔可夫链

Abstract: This paper investigates low latency Virtual Network Function(VNF) chain deployment, based on Non-Uniform Memory Access(NUMA) architecture. To improve the resource utilization, certain number of VNFs are reused. This paper formulates the VNF deployment problem, which is proved to be NP-hard, and proposes an iteration deployment algorithm MarVNFP based on Markov chain. Extensive simulations show that MarVNFP reduces nearly 40% data transmission delay compared with well-known MRP algorithm.

Key words: network function virtualization, Non-Uniform Memory Access(NUMA), resource scheduling, Markov processes