ITR: Internet Flows as First-Class Values: Support for Dynamic, Flexible Internet Services

互联网最近的变形-从仅仅是尽力而为的传输介质到开放的通信和计算基础设施，需要开发健壮的抽象，这些抽象便于其用于支持不断增加的应用数量，符合广泛接受的正确性标准，这些标准确保可验证的安全，公平，安全，对新的因特网应用程序和服务进行编程时，也遇到了30年前困扰编程系统的同样的组织原则的缺乏问题。研究人员认为，认识到网络流是组织互联网编程系统的核心抽象，这可能是关键的组织原则。具体地说，为了在现有的互联网基础设施的限制下快速试验和部署各种新服务，有必要采用一种更强大的模型来命名、创建、组合、共享和处理互联网流。为了实现这一目标，研究人员将互联网流提升到一流的价值，即，根据需要直接访问、采取行动和编制程序的实体。因此，开发互联网应用或服务的任务（例如，CDN代理或端系统多播）变成了编写适当的程序来操纵因特网流（沿着与应用相关的其它第一类值）的任务。为了能够指定和编译操纵互联网流的程序，特别是特定应用的互联网协议，有必要进行包括两个计算机科学学科的基础研究：网络系统和编程语言。需要对网络系统进行基础研究，以开发支持各种抽象的核心服务，并将其提供给协议程序员。我们需要对程式语言进行基础研究，以发展原语和形式化，使能表达协议的特性，以及验证协议的某些安全特性。建议的工作将集中于以下特定范畴：（1）支持各种程度的可靠性、拥塞管理和及时性的传输服务，（2）支持各种程度的移动性和多播的路由服务，（3）端到端服务，以支持不同程度的QoS和安全性，（4）可扩展机制，以支持因特网服务的灵活组合，以及（5）类型推断，以支持高效和安全的面向流编程。拟议中的研究将是一项合作努力，主要由网络系统和程序设计领域内已建立的研究小组的成员进行。拟议工作的组成部分是实施和原型制作。为此，将通过实现NetBench（一种编程环境，其中支持一组核心网络流类型和操作），沿着类型检查和类型推理系统来处理网络流类型，来证明互联网流是一流值的范例的效用。实现这些目标将超越目前旨在支持互联网增长的零敲碎打的尝试。研究人员认为，拟议的努力将提高网络软件开发环境的灵活性，可靠性，安全性和安全性，从而提高经济效率，类似于传统编程语言研究所带来的软件开发人员的更高生产力。为追求这些雄心勃勃的目标而组建的研究团队为编程语言和互联网技术的研究做出了重大的、全国公认的贡献，并在软件开发和技术转让方面取得了良好的记录。波士顿大学致力于通过大量的财政和基础设施承诺来支持这个团队，这些承诺补充和利用了从NSF寻求的支持。

The recent metamorphosis of the Internet-from a mere best-effort transport medium to an open communication and computation infrastructure necessitates the development of robust abstractions that facilitate its use to support a constantly increasing number of applications, in compliance with widely-accepted correctness standards that ensure a verifiably safe, fair, secure, and efficient access of Internet resources.Programming new Internet applications and services suffers from the same lack of organizing principles that afflicted programming systems some thirty years ago. The researchers believe that recognizing network flows as the central abstraction around which to organize a programming system for the Internet is perhaps the key organizing principle. Speciffically, to rapidly experiment with and deploy a wide range of new services within the existing constraints of the Internet infrastructure, it is necessary to adopt a more powerful model for the naming, creation, composition, sharing, and processing of Internet flows.Towards this goal the researchers promote Internet flows to first-class values, i.e., entities that are directly accessed, acted on and programmed according to needs. As a consequence, the task of developing an Internet application or service (e.g., a CDN broker or an end-system multicast) becomes the task of writing an appropriate program to manipulate Internet flows (along with other first-class values relevant to the application). To enable the speciffication and compilation of programs manipulating Internet flows, in particular application-specific Internet protocols, it is necessary to conduct basic research that encompasses two computer science disciplines: Networking Systems and programming Languages. Basic research in networking systems is needed to develop the core services that support the various abstractions to be presented to protocol programmers. Basic research in programming languages is needed to develop the primitives and formalisms that enable the expression of protocol properties and the verification of certain safety properties of the protocol.The proposed work will focus on the following specific areas: (1) Transport Services to support various degrees of reliability, congestion management, and timeliness, (2) Routing Services to support various degrees of mobility and multicasting, (3) End-to-End Services to support various degrees of QoS and security, (4) Naming mechanisms to support flexible composition of Internet services, and (5) Type inference to support efficient and safe flow-oriented programming.Towards the goals outlined above, the proposed research will be a collaborative effort carried primarily by members of established research groups in Networking Systems and in Programming Languages.A key component of the proposed work is implementation and prototyping. To that end, the utility of a paradigm in which Internet flows are first-class values will be demonstrated by implementing NetBench a programming environment in which a core set of network flow types and operations will be supported, along with a type-checking and type-inference system to handle types of network flows.The pursuit of the research goals outlined in this proposal is timely. Achieving these goals will leapfrog current piecemeal attempts aiming at supporting Internet growth. The researchers believe that the proposed effort will improve the flexibility, reliability, safety, and security of network software development environments, leading to economic efficiencies similar to the higher productivity of software developers arising from traditional programming language research. The research team assembled to pursue these ambitious goals has made significant, nationally-recognized contributions to research in Programming Languages and Internet Technologies and has an established record in software development and technology transfer. Boston University is committed to supporting this team through substantial financial and infrastructural commitments that complement and leverage the support sought from NSF.