网络功能虚拟化（NFV）采用商业服务器、交换机等通用设备实现多种网络功能，以替代原本采用专用设备的中间盒，但通用服务器和网络设备通常有较高的故障率。另一方面，NFV通过一系列新兴技术构建服务功能链（SFC），以较低的开销提供灵活、可扩展的服务，这些新技术也给NFV系统的健壮性带来了新的挑战。已有的NFV相关研究对健壮性的关注较少。本课题针对NFV服务功能链健壮性问题展开研究。(1)研究SFC路由健壮性,利用有限的资源在网络发生故障时对服务功能链的路由提供快速有效的保护。(2)研究SFC编排健壮性，通过高效的在线编排算法对可扩展性、资源利用率、健壮性等多个指标进行联合优化，与SFC路由保护相结合，减小故障时重建SFC的时间和资源开销。(3)研究SFC扩展和迁移健壮性，对扩展和迁移进行智能决策并对SFC迁移进行故障保护。开发模拟系统和原型系统，通过基于真实数据的实验验证所提方案的有效性。

Network function virtualization (NFV) realizes various network functions using general-purpose devices such as commercial servers and switches, so as to substitute traditional dedicated devices-based middleboxes. However, such general-purpose servers and network devices suffer from a high failure rate. On the other hand, a set of emerging technologies are developed to construct service function chains (SFCs) for NFV. Such technologies enable flexible and scalable services with little cost, but they also introduce new challenges to the robustness of NFV systems. This project focuses on robustness of NFV, which few existing studies have been considered. First, we study the robustness of SFC routing. We propose to protect SFC routing against failures by fast rerouting with constraints of limited resources. Second, we study the robustness of SFC orchestration. We jointly optimize scalability, resource utilization, and robustness by developing efficient online orchestration algorithms, so as to reduce time and resource cost when reconstructing SFC under failure. Third, we study the robustness of SFC scaling and migration. We focus on intelligent decision-making between SFC scaling and migration, and we propose to protect SFC migration against failures. We plan to develop simulation and prototype systems, and validate the efficacy of proposed schemes by comprehensive simulations and experiments based on real data traces.