基于VPP和创新性算法的高性能网络仿真平台构建

如今，在对新的网络技术进行研究时，研究人员们通常首先进行仿真验证，然后再将已验证的网络技术迁移到实际环境中。然而现有的网络仿真器主要用于网络配置的验证，很难用于执行高性能网络协议测试或数据中心网络压力测试等面向性能的任务。为了解决以上问题，本文构建了一个创新性的网络仿真器CNNet，旨在于仿真高吞吐量网络并在仿真网络上执行高性能需求的任务。CNNet遵从了云原生的设计思想，利用K8s作为云管理平台来管理集群资源，并保留了多种云管理平台的接口。CNNet还利用了以VPP为代表的高性能用户态数据平面，通过在服务器端及仿真节点端之间创建并配置Memif接口对及Tap接口对来构建高性能虚拟网络链路，并以此来进行不同仿真节点之间的通信。为了进一步提高仿真网络的性能，本文还提出了一种新的图分割算法以将仿真网络嵌入到底层物理服务器集群中。该算法同时达到了两个目标，即平衡每个工作端服务器上的仿真节点消耗的计算资源，以及最小化跨机虚拟链路的物理带宽消耗。最后，本文在一个服务器集群上评估了CNNet的性能，该集群由4台高性能服务器及1台普通服务器组成，其内部以100Gbps交换机进行高速连接。结果表明，CNNet可以在100个节点的胖树拓扑网络仿真实验下达到70Gbps的bi-section吞吐量，以及105个节点的叶脊拓扑网络仿真实验下达到96Gbps的bi-section吞吐量，这比分布式Mininet仿真器及其他主流的仿真平台的性能高出了一个数量级。本文构建了一个云原生化的网络仿真平台，并创新性提出了将高性能用户态数据平面与网络仿真平台相结合，以执行高性能网络的仿真任务。除此之外，本文还将提出了一个创新性的图分割算法以提高仿真平台的性能。

[1] 刘诚明. 中国联通研究院创新研究系列丛书: 软件定义网络技术与应用[M]. 北京: 人民邮电出版社, 2013.
[2] 约兰松. 经典译丛信息网络技术与网络科学: 软件定义网络原理、技术与实践[M]. 北京:电子工业出版社, 2016.
[3] LEE S, ALI J, ROH B H. Performance Comparison of Software Defined Networking Simulators for Tactical Network: Mininet vs. OPNET[C]//2019 International Conference on Computing, Networking and Communications (ICNC). 2019: 197-202.
[4] Wikimedia Foundation, Inc. Software-defined Networking[EB/OL]. 2023
[2023-02-24]. https://en.wikipedia.org/wiki/Software-defined_networking.
[5] MILLARD C. Cloud Computing Law[M]. Oxford: Oxford University Press, 2013.
[6] NoviFlow inc. The basics of SDN and the OpenFlow Network Architecture[EB/OL]. 2023
[2023-02-22]. https://noviflow.com/the-basics-of-sdn-and-the-openflow-network-architecture/.
[7] KATSIKAS G P, BARBETTE T, KOSTIć D, et al. Metron: High-Performance NFV Service Chaining Even in the Presence of Blackboxes[J]. ACM Trans. Comput. Syst., 2021, 38(1–2).
[8] 王进文, 张晓丽, 李琦, 等. 网络功能虚拟化技术研究进展[J]. 计算机学报, 2019, 42(2): 415-436.
[9] 周伟林, 杨芫, 徐明伟. 网络功能虚拟化技术研究综述[J]. 计算机研究与发展, 2018, 55(4): 675-688.
[10] YOUSAF F Z, BREDEL M, SCHALLER S, et al. NFV and SDN-Key Technology Enablers for 5G Networks[J]. IEEE journal on selected areas in communications, 2017, 35(11): 2468-2478.
[11] SHIRMARZ A, GHAFFARI A. Performance issues and solutions in SDN-based data center: a survey[J]. The Journal of supercomputing, 2020, 76(10): 7545-7593.
[12] MEDHI D, RAMASAMY K, ZUPAN J. The Morgan Kaufmann series in networking: Network Routing: Algorithms, Protocols, and Architectures[M]. San Francisco: Elsevier Science and Technology, 2010.
[13] IRI M. Mathematics in science and engineering: Network flow transportation and scheduling theory[M]. New York: Academic Press, 1969.
[14] 张高煜. 21 世纪高等学校计算机应用技术规划教材: 计算机网络技术实训[M]. 北京: 清华大学出版社, 2011.
[15] Mininet Project Contributors. Mininet[EB/OL]. 2022
[2022-06-21]. http://mininet.org/.
[16] NANAM, Inc. NS-3[EB/OL]. 2022
[2022-06-21]. https://www.nsnam.org/.
[17] LAI J, TIAN J, ZHANG K, et al. Network Emulation as a Service (NEaaS): Towards a CloudBased Network Emulation Platform[J]. Mobile networks and applications, 2020, 26(2): 766-780.54
[18] LIU H, ZHU Y, PADHYE J, et al. CrystalNet: Faithfully Emulating Large Production Networks[C]//SOSP ’17: Proceedings of the 26th Symposium on operating systems principles. ACM,2017: 599-613.
[19] DI LENA G, TOMASSILLI A, SAUCEZ D, et al. Distrinet: A Mininet Implementation for The Cloud[J]. Computer communication review, 2021, 51(1): 3-9.
[20] SUSI G, GARCES P, CRISTINI A, et al. FNS: An Event-driven Spiking Neural Network Simulator Based on The LIFL Neuron Model[A]. 2020.
[21] KANNAN P G, SOLTANI A, CHAN M C, et al. BNV: Enabling Scalable Network Experimentation Through Bare-metal Network Vi rtualization[C]//11th USENIX Workshop on Cyber Security Experimentation and Test (CSET 18). Baltimore, MD: USENIX Association, 2018.
[22] BOLLA R, BRUSCHI R, RANIERI A, et al. Analyzing and Optimizing the Linux Networking Stack[C]//Grid Enabled Remote Instrumentation. New York, NY: Springer US, 2009: 187-199.
[23] The Linux kernel Contributors. Linux kernel source tree[EB/OL]. 2022
[2022-06-30]. https://github.com/torvalds/linux.
[24] Fd.Io. VPP Documentation[EB/OL]. 2022
[2022-06-30]. https://wiki.fd.io/view/VPP.
[25] The BESS Project Contributors. BESS Source File[EB/OL]. 2023
[2023-02-24]. https://github.com/NetSys/bess.
[26] EMMERICH P, PUDELKO M, BAUER S, et al. User Space Network Drivers[C]//ANRW’18: ANRW 2018 - Proceedings of the 2018 Applied Networking Research Workshop. NEW YORK: ACM, 2018: 91-93.
[27] DPDK Project. DPDK Official Website[EB/OL]. 2022
[2022-06-31]. https://www.dpdk.org.
[28] DPDK Project. DPDK Source File Tree[EB/OL]. 2022
[2022-06-31]. https://github.com/DPDK/dpdk.
[29] 李俊武. 云计算网络珠玑[M]. 北京: 电子工业出版社, 2015.
[30] REN Q, ZHOU L, XU Z, et al. PacketUsher: Exploiting DPDK to Accelerate Compute intensive Packet Processing[J]. Computer Communications, 2020, 161: 324-333.
[31] 宋卫平, 沈磊, 佘文魁. 基于 DPDK 的虚拟化系统高性能网络模块的研究与实现[J]. 科技创新导报, 2020, 17(20): 125-133.
[32] CAO J, LIU Y, ZHOU Y, et al. TurboNet: Faithfully Emulating Networks With Programmable Switches[J]. IEEE/ACM transactions on networking, 2022, 30(3): 1-15.
[33] CHAN K Y, LEE S S. Design and Implementation of P4 Virtual Switches and P4 Virtual Networks[J]. Computer Communications, 2023, 199: 126-138.
[34] SMYTH D, SCOTT-HAYWARD S, CIONCA V, et al. SECAP Switch—Defeating Topology Poisoning Attacks Using P4 Data Planes[J]. Journal of network and systems management, 2023, 31(1): 28-.
[35] ALVAREZ-HORCAJO J, MARTINEZ-YELMO I, LOPEZ-PAJARES D, et al. A Hybrid SDN Switch Based on Standard P4 Code[J]. IEEE communications letters, 2021, 25(5): 1482-1485.
[36] PAOLUCCI F, SCANO D, CASTOLDI P, et al. Latency Control in Service Chaining Using P4-based Data Plane programmability[J]. Computer networks (Amsterdam, Netherlands : 1999), 2022, 216: 109227-. 55
[37] American City Business Journals. Intel is halting development of the networking chip it got from Barefoot Networks[EB/OL]. 2023
[2023-02-22]. https://www.bizjournals.com/sanjose/news/2023/01/26/intel-halts-development-of-tofino-switch-chips.html.
[38] STANTON I, KLIOT G. Streaming Graph Partitioning for Large Distributed Graphs[C]//KDD’12: Proceedings of the 18th ACM SIGKDD international conference on knowledge discovery and data mining. ACM, 2012: 1222-1230.
[39] TSOURAKAKIS C, GKANTSIDIS C, RADUNOVIC B, et al. FENNEL: Streaming Graph Partitioning for Massive Scale Graphs[C]//WSDM ’14: Proceedings of the 7th ACM international conference on web search and data mining. ACM, 2014: 333-342.
[40] XIE C, YAN L, LI W J, et al. Distributed Power-law Graph Computing: Theoretical and Empirical Analysis[C]//Advances in Neural Information Processing Systems: Vol. 2. 2014: 1673-1681.
[41] PETRONI F, QUERZONI L, DAUDJEE K, et al. HDRF: Stream-Based Partitioning for PowerLaw Graphs[C]//CIKM ’15: International Conference on Information and Knowledge Management, Proceedings. ACM, 2015: 243-252.
[42] BONAVENTURE O. OMNeT[J]. IEEE network, 2002, 16(4): 9-.
[43] CHEN M, MIAO Y, HUMAR I. OPNET IoT Simulation[M]. Singapore: Springer Singapore Pte. Limited, 2019.
[44] 塞西. 国际信息工程先进技术译丛: 计算机网络仿真 OPNET 实用指南[M]. 北京: 机械工业出版社, 2014.
[45] PEUSTER M, KARL H, VAN ROSSEM S. MeDICINE: Rapid Prototyping of Production ready Network Services in Multi-PoP Environments[C]//2016 IEEE Conference on Network Function Virtualization and Software Defined Networks (NFV-SDN). 2016: 148-153.
[46] PFAFF B, PETTIT J, KOPONEN T, et al. The Design and Implementation of Open vSwitch[C]//NSDI’15:12th USENIX symposium on networked systems design and implementation. 2015: 117-130.
[47] 张坤, 朱克俊, 黄秀添, 等. 浅析嵌入式 Linux 部署 Open vSwitch 及应用[J]. 有线电视技术, 2019(3): 80-86.
[48] 王文涛, 王奇枫, 郭峰, 等. 基于 Open vSwitch 的 SDN 网络平台构建方法[J]. 中南民族大学学报自然科学版, 2014, 33(4): 99-104.
[49] PU Y, DENG Y, NAKAO A. Cloud Rack: Enhanced Virtual Topology Migration Approach With Open vSwitch[C]//International Conference on Information Networking 2011, ICOIN 2011. IEEE, 2011: 160-164.
[50] PUJOLLE G. Networks and telecommunication series: Software networks : virtualization, SDN, 5G and security[M]. Revised and updated 2nd edition. ed. London: ISTE Ltd., 2020.
[51] BOSSHART P, DALY D, GIBB G, et al. P4: Programming Protocol-independent Packet Processors[J]. ACM SIGCOMM Computer Communication Review, 2014, 44(3): 87-95.
[52] MICHEL O, BIFULCO R, RéTVáRI G, et al. The Programmable Data Plane: Abstractions, Architectures, Algorithms, and Applications[J]. ACM computing surveys, 2021, 54(4): 1-36.56参考文献
[53] MADUREIRA A L R, ARAúJO F R C, SAMPAIO L N. On Supporting IoT Data Aggregation Through Programmable Data Planes[J]. Computer networks, 2020, 177: 107330-.
[54] ZHANG X, PAN H, XIE G. Progress in Programmable Network Data Plane[J]. Telecommunications Science, 2022, 38(6): 42-.
[55] ZHANG C, BI J, ZHOU Y, et al. HyperVDP: High-Performance Virtualization of the Programmable Data Plane[J]. IEEE journal on selected areas in communications, 2019, 37(3): 556-569.
[56] THOMPSON C D. A Complexity Theory For VLSI[M]. 1980-01-01.
[57] The Kubernetes Authors. Kubernetes Documentation[EB/OL]. 2022
[2022-06-30]. https://kubernetes.io/docs/home/.
[58] PONISZEWSKA-MARAńDA A, CZECHOWSKA E. Kubernetes Cluster for Automating Software Production Environment[J]. Sensors (Basel, Switzerland), 2021, 21(5): 1-24.
[59] LEE G, GIRI R A. Cloud Networking: Understanding Cloud-Based Data Center Networks[M]. San Francisco: Elsevier Science and Technology, 2014.
[60] 刘圣. VXLAN 技术在数据中心的应用[J]. 金融科技时代, 2018(11): 28-31.
[61] 张少芳, 刘延锋. 云计算背景下 VXLAN 技术的应用[J]. 无线互联科技, 2019, 16(9): 136-138.
[62] Fd.Io. VPP Source Tile Tree[EB/OL]. 2022
[2022-06-30]. https://github.com/FDio/vpp.
[63] Wikimedia Foundation, Inc. Tap introduction[EB/OL]. 2022
[2022-06-31]. https://github.com/FDio/vpp.
[64] DPDK Project. Memif Poll Mode Driver Documentation[EB/OL]. 2022
[2022-06-31]. https://doc.dpdk.org/guides/nics/memif.html.
[65] The Kubernetes Authors. Flannel CNI instruction[EB/OL]. 2023
[2023-02-24]. https://kubernetes.feisky.xyz/extension/network/flannel.
[66] The Kubernetes Authors. Calico CNI instruction[EB/OL]. 2023
[2023-02-24]. https://kubernetes.feisky.xyz/extension/network/calico.
[67] NELSON B J. Remote Procedure Call Technical Report CSL-81-9[J]. Xerox PARC, Palo Alto, Calif, 1981.
[68] Fd.Io. The Performance Ceasurement of Vector Packet Processor(VPP)[EB/OL]. 2023
[2023-02-24]. https://fd.io/docs/vpp/v2101/whatisvpp/performance.html.
[69] DUAN J, YI X, ZHAO S, et al. NFVactor: A Resilient NFV System Using the Distributed Actor Model[J]. IEEE journal on selected areas in communications, 2019, 37(3): 586-599.
[70] STALLINGS W. Foundations of modern networking : SDN, NFV, QoE, IoT, and Cloud[M]. Indianapolis, Indiana: Pearson, 2016 - 2016.
[71] ANDREEV K, RäCKE H. Balanced graph partitioning[J]. Theory Comput. Syst, 2006: 929-939.
[72] Feldmann A E. Fast Balanced Partitioning is Hard, Even on Grids and Trees[A]. 2011. arXiv: 1111.6745.57
[73] FORTNOW L. The Golden Ticket : P, NP, and The Search For the Impossible[M]. Princeton: Princeton University Press, 2013.
[74] BATENI M, BEHNEZHAD S, DERAKHSHAN M, et al. Affinity clustering: Hierarchical clustering at scale[C]//Advances in Neural Information Processing Systems. 2017: 6865-6875.
[75] AYDIN K, BATENI M, MIRROKNI V. Distributed Balanced Partitioning Via Linear Embedding[J]. Algorithms, 2019, 12(8): 162-.
[76] LARK R M. The Cambridge Dictionary of Statistics[J]. European journal of soil science, 2011, 62(2): 333-333.
[77] AL-FARES M, LOUKISSAS A, VAHDAT A. A Scalable, Commodity Data Center Network Architecture[J]. Computer communication review, 2008, 38(4): 63-74.
[78] OHTA S. The Number of Rearrangements for Clos networks –new results[J]. Theoretical computer science, 2020, 814: 106-119.
[79] CLOS C. A Study of Non-blocking Wwitching Networks[J]. The Bell System Technical Journal, 1953, 32(2): 406-424.
[80] Valter Popeskic. CLOS Topology[EB/OL]. 2023
[2023-02-24]. https://howdoesinternetwork.com/2019/clos-topology.
[81] GALLO M A. Networking explained[M]. 2nd ed. Boston: Digital Press, 2002.
[82] KOK C W, SALIM BEG M. Efficient Routing for IP Subnet VLAN Over Ethernet[J]. International journal of communication systems, 2002, 15(1): 67-83.
[83] 洪联系. 新世纪应用型高等教育网络专业系列规划教材: 网络设备互联技术[M]. 大连:大连理工大学出版社, 2013.
[84] CHENG X, LIU Z, NING Y, et al. Cyber-physical Network Architecture for Data Stream Provisioning in Complex Ecosystems[J]. European transactions on telecommunications, 2022, 33
[85] 陆锋. 数据中心网络的 Spine-Leaf 架构[J]. IT 经理世界, 2020, 23(2): 82-.