CAREER: Network Samaritans (NETSAM): Network-Assisted Service for the Internet

端到端论点是互联网架构的一个基本原则。然而，互联网的商业化带来了新的需求，如信任、服务质量、ISP差异化、第三方参与和不那么复杂的用户。在寻求一种新的互联网架构的过程中，这些需求促使人们重新思考端到端原则。过去的经验表明，新架构的开发可能是一个漫长的过程，充满了争论，并将从一个巨大的、往往是未知的解决方案空间中汲取灵感。本研究将调查解决方案空间的一部分，即网络辅助服务作为未来互联网体系结构可能组成部分的可行性和适用性。定义了一种称为Network Samaritans （NetSam）的网络援助体系结构来映射此类服务的框架。NetSam的动机是观察到许多新需求似乎需要网络中的新机制。与活动网络不同，NetSam是向增强网络迈出的温和而谨慎的一步，它提供了一组超出基本转发的服务，但在精神上与an的可伸缩性网络非常遥远。NetSam通过寻求确定网络中既必要又适当的最小功能，保留了Internet的体系结构精神。NetSam不是一个Internet架构，而仅仅是一个以网络辅助为核心的新服务框架，它可能成为未来架构的一部分。方法本研究将在三个重要应用的背景下进行：可靠组播、分布式拒绝服务和网络管理。调查将遵循一种严格的、极简的方法，假设除非证明不需要服务，否则不需要服务。选定的服务将被分解，其组件将分布在网络和应用程序之间。网络组件将是简单和通用的，因此它们可以很容易地在路由器中实现，安全风险最小。NetSam体系结构定义了一种新的服务体系结构划分，即过滤器和代理服务。过滤器服务非常简单，通常在路由器中实现，而代理服务更复杂，通常在边缘实现（但使用转发过滤器的一些帮助）。过滤器和代理一起提供了灵活性并支持大量的服务。这项研究将采用自上而下的方法，隔离网络援助的应用。将定义简单和一般服务，候选服务包括数据包标记，转向和采样，然后将设计适当的过滤器。最终的系统将通过实施和仿真进行评估。本研究提出了一种应用/网络交互的新范式。与当前的范式不同，要么完全分离，要么将两者结合，这项研究提出了网络和应用程序之间的知情交互。这项研究对科学的主要贡献将是支配这种相互作用的体系结构。对教育的影响本研究将使学生了解应用/网络交互的理论和实践问题。学生将面临挑战，深入思考应用程序和网络之间的区别，重点是理解端到端论证和主动网络等概念。学生将学习分析应用需求，并考虑路由器辅助服务的优点和缺点。学生将学习如何定义一个新的架构。项目工作将要求学生定义新服务以及分解和实现其组成原语。

The end-to-end argument is a fundamental principle of the Internet architecture. The commercializationof the Internet, however, has brought new requirements like trust, QoS, ISP differentiation, third partyinvolvement and less sophisticated users. These requirements have prompted the rethinking of the end-to-endprinciple in a quest for a new Internet architecture. Past experience indicates that the development of anew architecture will likely be a long process, rich with debate and will draw from a vast and often unchartedsolution space. This research will investigate part of the solution space, namely the viability and suitability of NetworkAssisted Services as a possible component of the future Internet architecture. A network-assistance archi-tecture called Network Samaritans (NetSam) is defined to map the framework for such services. NetSam is motivated by the observation that many new requirements seemingly require new mechanisms in the net-work. Unlike Active Networks, NetSam is a moderate, yet careful step towards an enhanced network that provides a set of services beyond basic forwarding, but spiritually very distant from the malleable network of AN. NetSam preserves the architectural spirit of the Internet by seeking to determine the minimum func-tionality that is both necessary and appropriate in the network. NetSam is not an Internet architecture, but merely a framework for new services pivoting on network assistance, which may become part of the future architecture.Approach The study will be carried out in the context of three important applications: Reliable Multicast, Distributed Denial of Service, and Network Management. The investigation will follow a disciplined, minimalist approach, assuming that a service is not needed unless proven otherwise. Selected services will be decom-posed and their components distributed between the network and the application. Network components willbe simple and general so they can be easily implemented in the routers with minimal security risks. The NetSam architecture defines a new architectural division of services, namely filter and surrogateservices. Filter services are very simple and typically implemented in the routers, while surrogate servicesare more complex and typically implemented at the edges (but use some help from forwarding filters).Together, filters and surrogates provide flexibility and support a large spectrum of services. The study will follow a top-down approach by isolating the applications of network assistance. Simpleand general services will be defined, with candidates including packet marking, steering and sampling andthen appropriate filters will be designed. The resulting system will be evaluated through implementation andsimulation.Impact on Advancing Knowledge This research proposes a new paradigm for application/network interaction. Unlike current paradigmsthat follow either complete separation or union of the two, this research proposes informed interactionbetween the network and application. The major contribution of this research to science will be the architecturethat governs this interaction.Impact on Education This research will expose students to both theoretical and practical issues on application/network interaction. Students will be challenged to think deeper about the distinction between the application and the network, with emphasis on understanding concepts like the end-to-end argument and active networking.Students will learn to analyze application requirements and consider the merits and demerits of router-assisted services. Students will learn how to define a new architecture. Project work will task students with the definition of new services and the decomposition and implementation of their constituent primitives.