Latency-optimized SDN in wide area networks

广域网中延迟优化的 SDN

• 批准号: 506203-2016
• 负责人: Khisti, Ashish
• 金额: $ 3.04万
• 依托单位: University of Toronto
• 依托单位国家: 加拿大
• 项目类别: Collaborative Research and Development Grants
• 财政年份: 2019
• 资助国家: 加拿大
• 起止时间: 2019-01-01 至 2020-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=688416
• 关键词: Latency; optimized; SDN; wide; area

Software Defined Wide Area Networks (SD-WAN) is one of the fastest growing markets in enterprise networks. By creating a logical overlay to intelligently route data across multiple paths, SD-WAN technologies provide higher capacities, lower costs and a simplified infrastructure to meet the ever increasing traffic demands. Many analysts predict that 2016 will be a breakout year for SD-WAN adaptation and implementation. The proposed project will make a fundamental contributions to SD-WANs by developing a new architecture to support applications with stringent latency constraints. Latency-optimization is a critical requirement for delay-sensitive applications such as interactive conferencing and desktop sharing, as well as cyber-physical systems and the Internet of Things. Such applications must compete with traffic generated from other sources including repair/maintenance applications or other delay tolerant applications in the backbone network. Furthermore in many real-world applications, packet retransmission have to be disabled and Forward Error Correction (FEC) must be used to avoid round-trip propagation delays. To address these challenges, the proposed research will develop a novel SD-WAN solution with integrated Forward Error Correction (FEC) for minimizing end-to-end latencies for delay sensitive applications. The proposed research will lead to the first fully functional FEC integrated SD-WAN that makes joint decisions on error correction and network path selection for latency optimization. On the theory side we will design new delay-sensitive FECs that can be seamlessly integrated into the SD-WAN configurations, as well as novel controller-algorithms for large scale networks with multiple flows. On the practical side we will implement software-based routers and high-performance controllers that can scale to heavy traffic volumes with tens of thousands of flows, perform query and update operations at line-speeds (Gbps), and use cross-layer network state information from the viewpoint of minimizing latency.

软件定义广域网（SD-WAN）是企业网络中增长最快的市场之一。通过创建逻辑覆盖以跨多个路径智能地路由数据，SD-WAN技术提供更高的容量，更低的成本和简化的基础设施，以满足不断增长的流量需求。许多分析师预测，2016年将是SD-WAN适应和实施的突破年。拟议的项目将通过开发一种新的架构来支持具有严格延迟限制的应用程序，为SD-WAN做出根本性的贡献。 延迟优化是交互式会议和桌面共享等延迟敏感型应用程序以及网络物理系统和物联网的关键要求。这样的应用必须与从其它源（包括骨干网络中的修理/维护应用或其它延迟容忍应用）生成的业务竞争。此外，在许多实际应用中，必须禁用数据包重传，并且必须使用前向纠错（FEC）来避免往返传播延迟。为了应对这些挑战，拟议的研究将开发一种新型的SD-WAN解决方案，该解决方案具有集成的前向纠错（FEC），以最大限度地减少延迟敏感应用的端到端延迟。 拟议的研究将导致第一个功能齐全的FEC集成SD-WAN，它可以对纠错和网络路径选择进行联合决策，以优化延迟。在理论方面，我们将设计新的延迟敏感FEC，可以无缝集成到SD-WAN配置中，以及用于具有多个流的大规模网络的新型路由算法。在实践方面，我们将实现基于软件的路由器和高性能控制器，这些路由器和控制器可以扩展到具有数万个流的大流量，以线速（Gbps）执行查询和更新操作，并从最小化延迟的角度使用跨层网络状态信息。