M. Kourtis, G. Xilouris, Vincenzo Riccobene, M. Mcgrath, G. Petralia, H. Koumaras, G. Gardikis, F. Liberal
{"title":"Enhancing VNF performance by exploiting SR-IOV and DPDK packet processing acceleration","authors":"M. Kourtis, G. Xilouris, Vincenzo Riccobene, M. Mcgrath, G. Petralia, H. Koumaras, G. Gardikis, F. Liberal","doi":"10.1109/NFV-SDN.2015.7387409","DOIUrl":null,"url":null,"abstract":"The primary goal of Network Function Virtualization (NFV) is the migration of physical network functions to software versions running on virtual machines (VM) in cloud computing environments. Rapid developments in virtualization technologies have made high-speed network connections and line rate packet processing viable in virtualized infrastructure environments. Deep Packet Inspection (DPI) of network traffic in the form of a computational intensive virtualized network function (VNF) was selected as a representative use-case. The DPI use case was used, to demonstrate the benefits of using SR-IOV enabled devices with DPDK to support performant Virtual Network Function (VNF) deployments. Performance evaluation of VNF versions using LibPCAP, SR-IOV and DPDK have been carried out. The results demonstrate that significantly higher packet throughput performance can be achieved when using SR-IOV and DPDK in unison in comparison to packet processing with the native Linux kernel network stack.","PeriodicalId":315251,"journal":{"name":"2015 IEEE Conference on Network Function Virtualization and Software Defined Network (NFV-SDN)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2015-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"62","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2015 IEEE Conference on Network Function Virtualization and Software Defined Network (NFV-SDN)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/NFV-SDN.2015.7387409","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 62
Abstract
The primary goal of Network Function Virtualization (NFV) is the migration of physical network functions to software versions running on virtual machines (VM) in cloud computing environments. Rapid developments in virtualization technologies have made high-speed network connections and line rate packet processing viable in virtualized infrastructure environments. Deep Packet Inspection (DPI) of network traffic in the form of a computational intensive virtualized network function (VNF) was selected as a representative use-case. The DPI use case was used, to demonstrate the benefits of using SR-IOV enabled devices with DPDK to support performant Virtual Network Function (VNF) deployments. Performance evaluation of VNF versions using LibPCAP, SR-IOV and DPDK have been carried out. The results demonstrate that significantly higher packet throughput performance can be achieved when using SR-IOV and DPDK in unison in comparison to packet processing with the native Linux kernel network stack.
网络功能虚拟化(Network Function Virtualization, NFV)的主要目标是将物理网络功能迁移到云计算环境中运行在虚拟机上的软件版本。虚拟化技术的快速发展使得高速网络连接和线路速率数据包处理在虚拟化基础设施环境中成为可能。选择计算密集型虚拟网络函数(VNF)形式的网络流量深度包检测(DPI)作为代表性用例。使用DPI用例来演示使用支持SR-IOV的设备和DPDK来支持高性能虚拟网络功能(VNF)部署的好处。利用LibPCAP、SR-IOV和DPDK对VNF版本进行了性能评估。结果表明,与使用本地Linux内核网络堆栈进行数据包处理相比,同时使用SR-IOV和DPDK可以实现更高的数据包吞吐量性能。