Virtual Network Function Instance Deployment and Performance Testing

Su Chen, Jingya Zhao, Mei Chen, Zhenyu Gao · 2022

Emerging concepts represented by “5G network”, “data center network” and “edge computing” all rely on the SDN/NFV framework. In the NFV framework, VNF or virtual network function exists as the smallest unit. Research has revealed the vital significance of studying the attributes of VNF instances to NFV network modeling. First of all, this paper realizes deployment of VNF instances in the Linux server, and makes use of DPDK optimization to obtain multiple independent VNF instances with favorable performance for follow-up research. Additionally, a TRex-based network performance test plan is designed and realized. According to this plan, various customized flows can be generated to achieve measurement of the throughput capacity and data packet handling latency of VNF instances. At last, cgroup is used to accomplish allocation of computing resources, which is then combined with the network performance test plan for an actual flow test of multiple VNF instances. Finally, performance data of VNF instances under different computing resource allocations can be obtained, and the VNF instance performance-resource characteristic curve is drawn.

Read the paper · More papers on PaperTik