CNS Core:Medium: NFLambda -- A Granular, Scalable and Secure NFV Framework for High Performance Packet Processing at 100 Gbps and Beyond

CNS 核心：中：NFLambda——一种精细、可扩展且安全的 NFV 框架，用于 100 Gbps 及以上的高性能数据包处理

• 批准号: 2106771
• 负责人: Zhi-Li Zhang
• 金额: $ 120万
• 依托单位: University of Minnesota-Twin Cities
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2021
• 资助国家: 美国
• 起止时间: 2021-10-01 至 2024-09-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2106771&HistoricalAwards=false
• 关键词: CNS; Core; Medium; NFLambda; Granular

The rapid growth of network bandwidth at the edge makes it possible to support high bandwidth 8K & volumetric video streaming, Augmented Reality/Virtual Reality and Distributed AI and IoT applications. The growing traffic demands from these new applications will pose enormous burdens on cellular packet core networks and mobile edge clouds, where NFV is meant to be a key enabling technology. The software nature of network function virtualization (NFV) enables network operators to dynamically scale out/in by instantiating more/fewer instances of network functions (NFs) in accordance with traffic demands or in preparation for or reaction to failures, thereby providing high scalability, availability and fault tolerance. Perhaps more importantly, NFV endows network carriers and service providers with the ability to quickly adapt, upgrade or roll out new network features or services. This proposal advances a new "greenfield" framework for refactoring and re-architecting NFV, referred to as NFLambda. The goal is to tackle the challenges in scaling packet processing for service function chaining in commodity multi-core servers to 100 Gbps and beyond. If successful, NFLamda will serve as an enabling technology for creating more cost effective and elastic service environments in emerging 5G and other access networks. The goal of this project is to tackle the challenges in scaling service function chain packet processing in commodity multi-core servers to 100 Gbps and beyond, while at the same time being able to fully take advantage of the software nature of NFV. NFLambda is designed with several salient features using a novel actor framework in the style of functional reactive programming (FRP). The NFLambda actor framework provides a powerful declarative, secure-by-design programming model with built-in monitoring and security mechanisms. The proposal advances a principled approach for decomposing conventional ("monolithic") NFs into a collection of granular actors via separation of control and data, refactoring both states and operations. NFLambda is designed to not only take into account the server architecture and cache/memory access resource constraints, but also effectively leverage software/hardware capabilities (e.g., compilation optimization techniques and hardware accelerators), smart NICs and programmable switches to achieve high performance packet processing. NFLambda is highly scalable, resilient and secure by design. The project will engage undergraduate, women and URM students in integrative research and education and provide outreach to K-12 students. The project will also engage industrial partners for tech transfer.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

边缘网络带宽的快速增长使得支持高带宽8 K容量视频流、增强现实/虚拟现实以及分布式AI和物联网应用成为可能。 这些新应用不断增长的流量需求将给蜂窝分组核心网络和移动的边缘云带来巨大的负担，而NFV是其中的关键技术。 网络功能虚拟化（NFV）的软件性质使得网络运营商能够通过根据业务需求或在对故障的准备或反应中实例化更多/更少的网络功能（NF）实例来动态地向外/向内扩展，从而提供高可扩展性、可用性和容错性。也许更重要的是，NFV赋予网络运营商和服务提供商快速适应、升级或推出新网络功能或服务的能力。 该提案提出了一种新的“格林菲尔德”框架，用于重构和重新架构NFV，称为NFLambda。目标是解决将商品多核服务器中服务功能链的数据包处理扩展到100 Gbps及以上的挑战。如果成功，NFLamda将作为一种支持技术，在新兴的5G和其他接入网络中创建更具成本效益和弹性的服务环境。该项目的目标是解决将商用多核服务器中的服务功能链数据包处理扩展到100 Gbps及以上的挑战，同时能够充分利用NFV的软件特性。NFLambda的设计具有几个突出的功能，使用了一个新颖的actor框架，采用了函数式反应式编程（FRP）的风格。NFLambda actor框架提供了一个强大的声明式、安全的编程模型，内置了监控和安全机制。该提案提出了一种原则性的方法，通过分离控制和数据，重构状态和操作，将传统的（“单片”）NF分解为一系列粒度行为体。NFLambda被设计为不仅考虑服务器架构和高速缓存/存储器访问资源约束，而且有效地利用软件/硬件能力（例如，编译优化技术和硬件加速器）、智能交换机和可编程交换机来实现高性能分组处理。NFLambda在设计上具有高度可扩展性、弹性和安全性。该项目将使本科生、妇女和URM学生参与综合研究和教育，并为K-12学生提供外展服务。该奖项反映了NSF的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

项目成果

Raven: belady-guided, predictive (deep) learning for in-memory and content caching

• DOI: 10.1145/3555050.3569134
• 发表时间: 2022-11
• 期刊: Proceedings of the 18th International Conference on emerging Networking EXperiments and Technologies
• 作者: Xinyue Hu;Eman Ramadan;Wei Ye;Feng Tian;Zhi-Li Zhang

PRAVEGA: Scaling Private 5G RAN via eBPF/XDP

PRAVEGA：通过 eBPF/XDP 扩展私有 5G RAN

• DOI: 10.1145/3609021.3609303
• 发表时间: 2023
• 期刊: ACM
• 作者: Dayalan, Udhaya Kumar;Wu, Ziyan;Gautam, Gaurav;Tian, Feng;Zhang, Zhi-Li
• 通讯作者: Zhang, Zhi-Li

NFlow and MVT Abstractions for NFV Scaling

用于 NFV 扩展的 NFlow 和 MVT 抽象

• DOI: 10.1109/infocom48880.2022.9796764
• 发表时间: 2022
• 期刊: IEEE INFOCOM 2022
• 作者: Wu, Ziyan;Zhang, Yang;Feng, Wendi;Zhang, Zhi-Li
• 通讯作者: Zhang, Zhi-Li

Accelerating Distributed Deep Learning using Multi-Path RDMA in Data Center Networks

• DOI: 10.1145/3482898.3483363
• 发表时间: 2021-10
• 期刊: Proceedings of the ACM SIGCOMM Symposium on SDN Research (SOSR)
• 作者: Feng Tian;Yang Zhang;Wei Ye;Cheng Jin;Ziyan Wu;Zhi-Li Zhang

Taproot: Resilient Diversity Routing with Bounded Latency

Taproot：具有有限延迟的弹性分集路由

• DOI: 10.1145/3482898.3483364
• 发表时间: 2021
• 期刊: ACM SIGCOMM Symposium on SDN Research (SOSR
• 作者: Ramadan, Eman;Mekky, Hesham;Jin, Cheng Jin;Dumba, Braulio;Zhang, Zhi-Li
• 通讯作者: Zhang, Zhi-Li