Optimizing general-purpose operating systems for high performance networking

长期以来，计算机操作系统一直为网络提供称为“Berkeley Sockets API”的编程接口。这起源于20世纪80年代中期的Unix操作系统的Berkeley软件分发(BSD)版本，此后在MacOS X、Windows和Linux中实现。因此，Berkeley Sockets API代表了一个重要的标准，并允许跨平台互操作性，但它在满足现代网络应用程序的需求方面进展缓慢。本研究将探索新的API、编程环境和操作系统功能如何支持现代和未来网络应用程序使用和部署高性能、灵活的网络传输协议。它将考虑网络性能相对于处理器速度的提高带来的挑战，以及由此产生的对低开销、零复制、并发网络协议堆栈的需要，以及支持日益灵活的传输协议的并行要求，并提供强大的安全和多路径支持。这项工作将特别考虑高级编程语言功能、新颖的网络传输协议和有效的操作系统抽象之间的相互作用；以及支持物联网和普适计算等新的新兴应用的能力。该项目建立在作为学生MSCI论文的一部分所做的先前工作的基础上，并且建立在互联网工程任务组(IETF)传输服务(TAPS)工作组正在进行的工作的基础上，以开发一种用于“POST Sockets API”的标准，但与该工作不同，并且采取了更具前瞻性的方法，因为标准工作必须更严格地考虑到向后兼容性的限制。工作涉及“体系结构和操作系统”、“ICT网络和分布式系统”、和“普适和无处不在的计算”(EPSRC的增长领域)。具体地说，它专注于如何最好地支持新出现的和未来的应用，包括移动、无处不在和物联网场景，并使其编程模型民主化，旨在更容易地利用网络的高级功能来支持这些应用。

Computer operating systems have long provided a programming interface to the network known as the "Berkeley Sockets API". This originated in the mid-1980s in the Berkeley Software Distribution (BSD) version of the Unix Operating System, and has since been implemented in MacOS X, Windows, and Linux. The Berkeley Sockets API thus represents an important standard and allows cross-platform interoperability, but it has been slow to evolve to meet the needs of modern networked applications.This research will explore how new APIs, programming environments, and operating systems features can support the use and deployment of high performance, flexible, network transport protocols for modern and future networked applications. It will consider the challenges posed by the increase in network performance relative to processor speed, and the resulting need to low-overhead, zero copy, concurrent network protocols stacks and the parallel requirements to support increasingly flexible transport protocols, with strong security and multipath support. The work will especially consider the interplay between advanced programming language features, novel network transport protocols, and efficient operating systems abstractions; and the ability to support novel and emerging applications such as the Internet of Things and pervasive computing.The project builds on prior work done as part of the student's MSci dissertation, and on ongoing work in the Internet Engineering Task Force (IETF) Transport Services (TAPS) working group to develop a standard for a "Post Sockets API", but is distinct from that work, and takes a more forward-looking approach since the standard work must account for thee limitations of backwards compatibility more strictly.The work relates to the "Architectures and operating systems", "ICT networks and distributed systems", and "Pervasive and ubiquitous computing" (an EPSRC growth area). Specifically, it focusses on how best to support novel emerging and future applications, including mobile, ubiquitous, and IoT scenarios, and to democratise theirprogramming model, aiming to make it easier to make use of the advanced features of the network to support these applications.