SDN新型网络中虚拟资源管理弹性架构及关键技术研究

Software Defined Network has a lot of interest in the research and the academic communities. The SDN concept evolves current design philosophies and could be assumed as one approach of a future clean slate Internet design. These main characteristics of software and virtualization advocated by SDN have now become an important trend in the evolution of future network. In these context, dynamic provisioning of customized network infrastructure as well as resource virtualization are appealing technologies. While the resource management model in current SDN architecture can't effectively ensures the efficient use of underlying network resources and the on-demand services quality of the upper business. To solve this problem, the project focuses on the flexible resource management architecture and its key technologies in virtualized environment. Firstly we propose an efficient and scalable management framework that enables end-to-end connectivity across heterogeneous virtual networks without revoking their autonomy, which builds a flexible, security and manageable virtualization environment. Then we study a language for virtual resources interconnection networks specification and modeling. Besides allowing end resources description, this language lets users describe the desirable virtual network topology, including virtual routers and timeline. We argue that the problems of the node mapping and the link mapping phases in the existing algorithms considerably reduce the solution space and degrade embedding quality. We design a novel virtual resources mapping algorithm based on the statistical analysis method of network topology. It can balance the network load and improve resource utilization. In order to improve the ability of Internet service and ensure the end-to-end reliable transmission, a virtual resource allocation framework based on cognitive techniques has been established by achieving self-optimization and reconfiguration. Finally, we set up a virtual resource mapping simulation platform and an intelligent management verification system to validate the effectiveness of the architecture and these methods. Extensive simulation experiments show that the proposed algorithms significantly outperform the existing heuristics by increasing the acceptance ratio and the revenue while decreasing the cost incurred by the substrate network. Through the experiments, we have successfully verified the resource management mechanism and network control functionalities. These details involved in this project are the emerging hot issues of communication. The solution of these problems can effectively promote the quality of services with diversity business in the innovation network environment and have realistic significance.

SDN所倡导的软件化和虚拟化，目前已经成为未来网络演进发展的重要趋势和主要特征，但当前SDN新型网络架构中资源管理模式无法有效保障底层网络资源的高效利用和上层业务的按需服务质量。为了解决该问题，本项目拟开展虚拟资源管理的弹性架构及其关键技术的研究。首先提出一个高效率、可扩展的虚拟资源管理弹性架构，搭建灵活、安全和可管理的端到端虚拟环境，研究虚拟资源统一描述语言，有效屏蔽底层网络的异构性。基于复杂网络的拓扑统计分析，设计拓扑感知的虚拟资源映射算法，平衡网络负载，提高资源利用率。建立基于认知技术的虚拟资源优化管理框架，实现智能调度与配置，满足差异化业务的QoS需求，保障端到端的可靠传输。最后搭建虚拟资源映射仿真平台和智能管理验证系统，对上述架构和方法进行测试验证。本项目所涉及的内容是通信领域的新兴热点问题，这些问题的解决能有效改善和提升创新网络环境下互联网服务质量，具有非常重要的现实意义。

SDN（软件定义网络）/NFV（网络功能虚拟化）已经成为未来网络创新的重要推动力量和主流趋势，成为全球主流厂商及运营商的战略重点。本项目基于SDN/NFV等颠覆性技术，开展了基于SDN的网络资源优化策略技术、基于SDN的虚拟服务迁移机制和基于SDN的虚拟数据中心的资源分配与管理关键技术的研究。通过本项目的支持，发表学术论文15篇，其中SCI 2篇，EI 12篇，申请专利4项，授权专利1项，授权软件著作权2项。具体的研究成果如下：.1、研究业务感知的网络优化策略关键技术，建立网络资源多维属性感知模型，设计基于SDN的策略优化控制算法。基于该研究成果，针对中移动现网的视频类业务，在浙江省、江西省和上海市的移动现网中搭建了跨越CMNET骨干网和省级专网的互联网内容流量多路径应急调度试验系统，对SDN内容分发核心功能内容路由调度及流量优化进行了试验。.2、基于SDN控制器感知迁移前后的源、目的位置，将源交换机上的相关策略转移到目的交换机上，实现网络策略的同步迁移。基于该成果的研究内容，在中移动私有云信息港内，选取PushMail邮件业务系统进行迁移实验的验证，通过服务器整合和业务系统的测试验证，实现用最小规模的虚机承载业务系统的正常运行，极大的提高了物理设施的利用率，降低了数据中心内的能耗。.3、将SDN与数据中心结合，提出基于拓扑势来评价网络中节点的重要性，指导VDC自动分配虚拟资源，提高数据中心的资源利用率，减小VDC映射的失败率。基于该成果的研究内容，发表多篇学术论文。MilCom的评审专家一致认为该问题能够为云计算和数据中心带来很大的利益，是一个非常有趣且时效性的问题（an interesting and timely problem）。.通过本项目的支持，项目负责人成功申请中国科学院声学研究所知识创新工程青年人才领域前沿项目《基于SDN实现智能管道网络资源优化策略技术研究》（2014-2015，8万），并参中国移动运营商开展深度合作，成功中标中移动“内容网络系统一期工程——内容管理平台”（2016.4-2017.4，182万）。