基于软件定义价格的SDN资源管理机制研究

发布时间：2018-03-10 13:31

本文选题：SDN　切入点：OpenDaylight　出处：《浙江工商大学》2014年硕士论文　论文类型：学位论文

【摘要】：如今,信息网络正面临着种种新的挑战,具备较强适应能力的软件定义网络正成为人们关注的焦点。软件定义网络(Software Defined Network, SDN)最初是由斯坦福大学clean slate研究组提出的一种新型网络架构,其核心技术是将网络设备控制面与数据面分离,通过对控制件的直接编程来实现灵活控制网络流量的目的。OpenDaylight是软件定义网络目前最有希望的一个开源框架,通过开放社区开发者、项目管理与开源代码的联合已大体实现了建立一个开放、可编程和可控制的参考框架。因此,本文结合经济学中双边市场多归属结构的概念,在OpenDaylight架构基础之上,提出了基于软件定义价格的SDN资源管理机制,并对该机制所涉及的相关技术做了更深入的探讨和研究。首先,本文详细介绍了软件定义网络的研究现状和未来发展趋势,并对软件定义网络技术进行分析对比。其次,在开源项目OpenDaylight基础之上提出了Multi-Agent价格协商模块,并将此模块嵌入到OpenDaylight架构中从而提出了基于Multi-Agent的OpenDaylight主体框架。再次,对经济学中的双边市场理论进行交叉学科研究,通过基于Multi-Agent的OpenDaylight主体框架采用双边市场多归属结构来构建SDN资源管理体系结构,并在该体系结构中提出了基于软件定义价格的SDN资源管理机制和相应的资源选择过程。然后,在该SDN资源管理机制中,本文提出来了软件定义价格(Software Defined Price,SDP)的概念。在SDP中,本文结合Mutil-Agent模型对资源进行定价,在博弈论的基础之上提出了一种动态的资源价格选择算法。在资源的分配上,分别在单、双平台上采用顺序分配策略和贪心策略进行分配。最后,本文通过不同资源分配策略在单平台和双平台下的仿真比较结果,可以得出以下结论：在任务完成代价和时间上,无论采用顺序分配策略还是贪心策略,双平台运行结果都比单平台更有优势,由此说明了基于软件定义价格的SDN资源管理体系结构的优越性。
[Abstract]:Today, the information network is facing new challenges. Software-defined networks with strong adaptability are becoming the focus of attention. Software Defined Networks (SDNs) is a new network architecture originally proposed by the clean slate team at Stanford University. The core technology is to separate the control surface of the network device from the data surface, and to realize the purpose of flexibly controlling the network traffic through direct programming of the control parts. OpenDaylight is the most promising open source framework for the software to define the network at present. Through open community developers, the combination of project management and open source code has largely achieved the establishment of an open, programmable and controllable reference framework. On the basis of OpenDaylight architecture, a SDN resource management mechanism based on software definition price is proposed, and the related technologies involved in this mechanism are discussed and studied deeply. First of all, this paper introduces the research status and future development trend of software definition network in detail, and analyzes and compares the software definition network technology. Secondly, based on the open source project OpenDaylight, a Multi-Agent price negotiation module is proposed. And the module is embedded into the OpenDaylight framework to put forward the main frame of OpenDaylight based on Multi-Agent. Thirdly, the cross-disciplinary research on the bilateral market theory in economics is carried out. Through the OpenDaylight principal framework based on Multi-Agent, the SDN resource management architecture is constructed by adopting the bilateral market multi-home structure, and the SDN resource management mechanism based on the software definition price and the corresponding resource selection process are put forward in this architecture. In this SDN resource management mechanism, this paper puts forward the concept of software definition price Defined (Defined). In SDP, this paper uses Mutil-Agent model to price resources. On the basis of game theory, a dynamic resource price selection algorithm is proposed, in which sequential and greedy strategies are used for resource allocation on single and double platforms, respectively. Finally, by comparing the simulation results of different resource allocation strategies on single and double platforms, we can draw the following conclusions: in terms of task completion cost and time, either sequential allocation strategy or greedy strategy is adopted. The results of dual-platform operation are superior to that of single platform, which shows the superiority of SDN resource management architecture based on software defined price.
【学位授予单位】：浙江工商大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.09

【参考文献】