NSF-BSF: NeTS: Small: Making BGP work for real-time interactive applications

几十年来，边界网关协议（BGP）一直是互联网的粘合剂，它决定了互联网流量从源到目的的路由。虽然它支持了Internet的巨大增长，但是Internet已经发生了变化，在其上运行的应用程序与Internet最初支持的应用程序非常不同。BGP不是性能感知的，所以它可能不会选择性能最好的路由。先前的研究发现，对于传统的互联网应用程序，如网页浏览、电子邮件和流媒体视频，它选择的路由通常足够好，可以容忍足够好的性能和暂时的波动。然而，远程健康监控、远程医疗、虚拟现实/自适应现实、自动驾驶汽车和机器人工厂等新兴用例需要可靠的低延迟、高吞吐量性能，以支持实时、交互式，有时甚至是关键的应用程序。尽管有这些需求，但BGP不能简单地替换或升级，因为成千上万的独立网络依赖它来互联和交换流量。该项目的目标有两个：(1)了解BGP的性能和可用性在什么情况下“足够好”，并确定什么情况下不是这样；(2)提出增强BGP性能和可用性的系统，而不需要对BGP本身进行不可部署的更改。如果成功，该项目将成为未来互联网服务的基础，在丰富的应用程序上提供可靠的实时交互，为世界各地的用户提供新的功能。该项目将研究BGP的性能和可用性，并探索BGP是否以及如何用于优化下一代实时基于互联网的服务的性能。该研究将在现代Internet应用程序交付中的三个关键挑战的背景下解决这些问题：确定将用户引导到哪个站点，跨广域路由到站点，以及选择从站点返回客户端的路由。它将开发使用BGP的新方法，使其能够作为新兴关键应用的基础。该项目将结合理论分析和与内容和云提供商长期合作的测量结果，并使用部署在真实互联网上的系统原型评估结果。该奖项反映了美国国家科学基金会的法定使命，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。

For decades, the Border Gateway Protocol (BGP) has served as the glue of the Internet, determining the routes that Internet traffic takes from source to destination. While it has supported the tremendous growth of the Internet, the Internet has changed and the applications that run on it are very different from those the Internet originally supported. BGP is not performance-aware, and so it may not choose the best-performing routes. Previous work found that the routes it chooses are often good enough for traditional Internet applications like web browsing, email, and streaming video that can tolerate good-enough performance and temporary blips. However, emerging use cases such as remote health monitoring, telemedicine, virtual reality/adaptive reality, self-driving cars, and robotized factories require reliable low-latency, high-throughput performance to support real-time, interactive, and sometimes critical applications. Despite these needs, BGP cannot simply be replaced or upgraded, as tens of thousands of independent networks rely on it to interconnect and exchange traffic.The project's objectives are twofold: (1) understanding the circumstances in which BGP’s performance and availability are "good enough" and identifying when this is not so; and (2) proposing systems for enhancing BGP’s performance and availability, without requiring undeployable changes to BGP itself. If successful, the project will be the basis for the Internet services of tomorrow, with dependable real-time interactions over rich applications, providing new capabilities to users around the world.The project will investigate the performance and availability of BGP and explore whether and how BGP can be used to optimize performance for the next generation of real-time Internet-based services. The research will address these questions in the context of three critical challenges in modern Internet application delivery: determining to which site to direct the user, routing across the wide area to the site, and selecting a route back to the client from the site. It will develop novel approaches to using BGP that enable it to serve as the underpinning of emerging critical applications. The project will combine theoretical analyses with measurements from long-standing collaborations with content and cloud providers and evaluate the results using system prototypes deployed on the real Internet.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.