SDN虚拟网络重映射算法设计及ONOS分布式数据库研究

发布时间：2018-05-02 02:12

本文选题：SDN + 网络虚拟化　；参考：《电子科技大学》2017年硕士论文

【摘要】：随着网络业务的发展,传统网络让网络运维管理变得复杂,新型网络架构SDN的出现打破了网络的封闭架构,提升了网络的灵活性。SDN架构实现了控制平面和数据转发平面的分离,有利于对网络业务进行集中控制,其灵活的编程方式也能够为实现网络虚拟化提供良好的平台,并且对未来网络的发展方向有很好的启迪作用。网络虚拟化是解决目前网络僵化的有效途径,该技术屏蔽了底层物理网络的复杂性,仅向上层提供租户所需的资源,降低了网络的运营成本,并对SDN网络架构有很好的支持。在网络虚拟化中,虚拟网络重配置有利于充分地利用底层资源,增加网络的收益,有效的重映射算法是实现该目的的关键。本文将SDN虚拟网络重映射算法作为研究课题,在SDN的背景下结合网络虚拟化,主要研究内容分为以下几个部分。首先,介绍了SDN和ONOS控制器的基本架构及其中的关键技术,分析了网络虚拟化技术的研究现状,并介绍了常见的虚拟化方式和网络功能虚拟化。其次,通过对重映射问题进行描述,设计了基于预测反馈的最小代价虚拟网络重映射算法。其中包括了基于同构子图搜索的虚拟网络映射算法和基于预测反馈的最小代价重配置算法两部分。使得系统能保持负载均衡,同时尽量减少重配置的代价以接收更多的虚拟网络,提高网络收益,并在ONOS系统中实现了设计的算法。然后,对ONOS分布式数据库进行深入研究。包括其底层协议的实现、数据查询方式和效率、分布式模式及对数据库事务的处理情况,并结合ONOS基于意图的策略,在集群工作状态下设计了批量计算任务进行相应的测试和结果分析。根据ONOS的分布式特性,通过在集群环境下迫使控制器下线对其可靠性和效率进行验证。最后,通过网络拓扑生成工具搭建实验环境对设计的算法进行仿真,并与已有算法进行对比分析得出结论。实验结果表明,虚拟网络重映射算法在网络代价和网络接收率上均有一定的改善。
[Abstract]:With the development of network services, the traditional network makes network operation and maintenance management complex. The emergence of a new network architecture, SDN, breaks the closed network architecture, and improves the flexibility of the network. SDN architecture realizes the separation of control plane and data forwarding plane. It is conducive to centralized control of network services, its flexible programming mode can also provide a good platform for the realization of network virtualization, and has a good enlightenment to the future development direction of the network. Network virtualization is an effective way to solve the current network fossilization. This technology shields the complexity of the underlying physical network, provides only the resources needed by tenants to the upper layer, reduces the operating cost of the network, and has a good support for the SDN network architecture. In network virtualization, virtual network reconfiguration is conducive to making full use of the underlying resources and increasing the network income. An effective remapping algorithm is the key to achieve this goal. In this paper, the SDN virtual network remapping algorithm is taken as the research topic, and the research content is divided into the following parts under the background of SDN combined with network virtualization. Firstly, the basic architecture of SDN and ONOS controllers and their key technologies are introduced. The research status of network virtualization technology is analyzed, and the common virtualization methods and network function virtualization are introduced. Secondly, by describing the remapping problem, a minimum cost virtual network remapping algorithm based on predictive feedback is designed. It includes two parts: the virtual network mapping algorithm based on isomorphism subgraph search and the minimum cost reconfiguration algorithm based on predictive feedback. The system can maintain load balance and reduce the cost of reconfiguration to receive more virtual networks and increase network income. The algorithm is implemented in ONOS system. Then, the ONOS distributed database is deeply studied. It includes the implementation of the underlying protocol, the way and efficiency of data query, the distributed mode and the processing of database transactions, and the strategy of ONOS based on intention. The batch computing task is designed to test and analyze the results in the working state of the cluster. According to the distributed nature of ONOS, the reliability and efficiency of the controller are verified by forcing the controller offline in the cluster environment. Finally, the experiment environment is built by the network topology generation tool, and the algorithm is simulated, and compared with the existing algorithms, the conclusion is drawn. The experimental results show that the virtual network remapping algorithm improves both the network cost and the network reception rate.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP393.0;TP311.13

【参考文献】