NeTS: Small: Scaling Routing: From Theory to Practice (and Back Again)

传统上，网络通过层次结构扩展到非常大的规模：在更高级别的聚合上执行路由，然后以更细的粒度执行。其后果是路径长度膨胀形式的低效，以及使用位置相关地址使管理、移动性和多宿主复杂化。基于与位置无关的“平面名称”的路由可以解决这些问题，并且对于支持安全的、自我认证的标识符是独立有用的。但之前提出的协议不存在用于直接在平面名称上进行可扩展、高效的路由。该项目正在开发分布式紧凑路由(DCR)协议，该协议在平面名称上进行路由，同时保证任意网络的可扩展性和效率。在紧凑型路由领域最新理论进展的基础上，DCR解决了包括自主分布式操作、拥塞和异构性在内的关键挑战。该项目正在将这些技术应用于互联网路由、移动自组织网络和以内容为中心的网络。此外，该项目正在研究最新的面向系统的建议，即基于将互联网的“核心”与边缘分开来扩展互联网路由，通过确定它们何时可以在几乎不损失网络效率的情况下提高可扩展性来提供坚实的理论基础。该项目的成果将显著提高互联网和其他网络的可扩展性、移动性和可管理性，弥合有希望的理论成果与新出现的实际需求之间的差距。结果将通过出版物和演示文稿、公开软件发布以及研究生和本科生课程来传播。

Traditionally, networks are scaled to very large size through hierarchy: routing is performed on higher level aggregates and subsequently at finer granularity. The consequences are inefficiency in the form of inflated path lengths, and the use of location-dependent addresses which complicate management, mobility, and multihoming. Routing on location-independent "flat names" can solve these problems and is independently useful to support secure, self-certifying identifiers. But no previously proposed protocols exist for scalable, efficient routing directly on flat names. This project is developing distributed compact routing (DCR) protocols which route on flat names while guaranteeing scalability and efficiency for arbitrary networks. Building on recent theoretical advances in the area of compact routing, DCR addresses key challenges including autonomous distributed operation, congestion, and heterogeneity. The project is applying these techniques to Internet routing, mobile ad hoc networks, and content-centric networks. In addition, the project is studying recent systems-oriented proposals to scale Internet routing based on separating the "core" of the Internet from the edges, providing firm theoretical footing by determining when they can improve scalability with little loss of network efficiency. The results of this project will significantly improve scalability, mobility, and manageability in the Internet and other networks, bridging a gap between promising theoretical results and emerging practical needs. Results will be disseminated through publications and presentations, a public software release, and graduate and undergraduate courses.