面向下一代网络的可编程测量架构及关键测量方法

网络测量为网络管理提供必需的输入信息，是网络管理不可或缺的组成部分。然而，现有网络在架构上缺乏对网络测量功能的支持，导致网络测量任务部署非常复杂，测量准确性、及时性和可扩展性差。作为新一代网络架构，可编程网络具有可编程控制、全局信息收集和数据包自定义处理能力，为更好地实施各种网络测量提供了技术支撑。本项目利用可编程下一代网络的可编程控制能力，研究灵活、高效和可扩展的可编程测量架构，在数据平面提供功能完备和灵活的网络状态测量能力，在控制平面上为各种测量任务提供简单灵活的抽象编程接口、分配测量资源和收集测量状态，并利用可编程下一代网络的灵活控制和自定义数据处理能力，研究准确、及时和可扩展的网络流量和性能指标测量方法。 在项目中，着重解决可编程测量架构和关键测量方法的灵活性、可扩展性和准确性问题，以及测量架构和方法中涉及的测量资源部署和分配问题。

Network measurement provides essential input information for network management tasks, and thus it plays a necessary role in network management. However, due to lack of network measurement function consideration in today's network architecture, network measurement in today's networks faces many issues, such as high deployment complexity, low measurement accuracy, low timeliness, and low scalability. As an emerging network architecture, programmable networking has several attractive features, such as programmable control plane, global network view and user-defined packet processing capability, which pave the way for networks to implement accurate, timely and scalable network measurement. In this project, we mainly study the following problems in programmable network measurement. First, by utilizing the programmable control capability of programmable networking, we aim to design an efficient programmable measurement architecture, which has powerful and flexible measurement capability in data plane, and provides flexible programming interfaces for measurement tasks and manages measurement resources and statistics collection in control plane. Second, by utilizing the flexible control and user-defined packet processing capability of programmable networks, we study the accurate, timely and scalable flow and network performance metric measurement schemes. Most importantly, we aim to solve the flexibility, scalability and accuracy issues as well as measurement resource deployment and allocation problems involved in the programmable measurement architecture and measurement schemes.