面向虚拟网调度的新型业务支撑系统设计与实现

发布时间：2018-05-30 08:10

本文选题：软件定义网络 + OpenFlow　；参考：《北京邮电大学》2016年硕士论文

【摘要】：随着云计算的蓬勃发展,各个云平台功能日趋丰富,同时SDN概念的出现与发展为网络虚拟化的不断创新提供了不可忽视的作用。SDN将控制逻辑从转发硬件中分离出来,便于网络的虚拟化和管理,从而能够简化新型协议和应用的部署,并且能够把各种网络中间件整合到软件实现的控制中。在此基础上,针对不同的网络结构,不断有新的网络调度策略被提出,从而满足不同的需求。而作为云平台的租户,目前很难对调度策略做自定义的配置,只能通过手动配置路由表项达到目的,如果云平台能够将不同网络调度策略提供给租户选择,就可以简化租户的配置,达到优化网络的目的。本文设计和实现了一个面向虚拟网调度的新型业务支撑系统,旨在研发一个可扩展、易修改、能移植的虚拟网调度平台,作为连接调度策略研究者与云平台租户的桥梁。使得租户可以在系统里选择满足自身网络需求的调度策略,优化私有网络。使得调度策略的研究开发人员可以在系统上发布自已的调度策略,同时,系统还预置部分调度策略,包括路径决策和流量分配。此外,通过对SDN网络调度模型的分析,本文结合蝙蝠算法和资源预留的思想提出了预测分配策略。在实现中,模块间采用REST API方式解耦,以组件的方式构建系统,方便修改实现或者替换组件。论文首先对系统的研究背景、国内外的研究现状以及系统所涉及的SDN相关技术做了简单介绍;在此基础上,提出了系统的具体需求,明确了系统需要实现的相关功能;然后,专门对SDN网络调度进行了分析,提出了预测分配的调度策略;接着对系统总体的设计、模块间的交互和各个接口做了概括性地介绍;再进一步详细地介绍了系统的各个关键模块的设计与实现;最后对系统的核心功能进行了测试,并对测试结果进行了分析,进而总结全文并提出下一步值得改进的地方。
[Abstract]:With the rapid development of cloud computing, the functions of each cloud platform are becoming more and more abundant. At the same time, the emergence and development of the concept of SDN provides an important role for the innovation of network virtualization. SDN separates the control logic from the forwarding hardware. It is convenient for network virtualization and management, which can simplify the deployment of new protocols and applications, and integrate all kinds of network middleware into the control of software implementation. On this basis, for different network structures, new network scheduling policies are proposed to meet different requirements. As the tenants of cloud platform, it is difficult to configure the scheduling policy, and it can only be achieved by manually configuring the routing table items. If the cloud platform can provide different network scheduling policies to tenants to choose, It can simplify the configuration of tenants and achieve the purpose of optimizing the network. This paper designs and implements a new service support system for virtual network scheduling, which aims to develop a scalable, easily modified, portable virtual network scheduling platform, which serves as a bridge between scheduling policy researchers and cloud platform tenants. So that tenants can choose a scheduling strategy to meet their network requirements in the system, and optimize the private network. It enables developers to publish their own scheduling policies on the system. At the same time, the system also presets some scheduling policies, including path decision and traffic allocation. In addition, based on the analysis of SDN network scheduling model, this paper proposes a predictive allocation strategy based on bat algorithm and resource reservation. In the implementation, the modules are decoupled by REST API, and the system is constructed by components, which is convenient to modify or replace the components. Firstly, the paper introduces the background of the system, the current situation of the research at home and abroad, and the related technologies of the system. On this basis, the specific requirements of the system are put forward, and the relevant functions of the system are defined. The SDN network scheduling is analyzed and the scheduling strategy of predictive allocation is put forward, then the overall design of the system, the interaction between modules and various interfaces are introduced in a general way. Furthermore, the design and implementation of the key modules of the system are introduced in detail. Finally, the core functions of the system are tested, and the test results are analyzed.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP311.52;TP393.09

【参考文献】