数据中心网络中服务功能链的部署与路由

发布时间：2018-03-19 08:24

本文选题：数据中心　切入点：服务功能体　出处：《中国科学技术大学》2017年硕士论文　论文类型：学位论文

【摘要】：当今数据中心部署了大量的服务器和网络设备,用以提供多种多样的网络服务功能(防火墙、内容缓存、广域网加速器等)和充分的网络连接。数据中心中大多数流量为多个服务功能体所处理,从而构成了传统网络服务链。传统网络服务链存在者一些缺点如网络设备之间的耦合性大、数据包转发效率低下、网络设备无法池化和扩展性差等。网络功能虚拟化(Network Function Virtualization,NFV)使得传统以硬件设备存在的网络功能可以以软件的形式大量部署在标准服务器上,服务功能体(Service Function,SF)可以动态创建,可灵活扩展;软件定义网络(Software Defined Networking,SDN)可以通过集中控制,使得控制器生成能够自动形成业务功能链的流表。基于SDN和NFV的功能链被称作服务功能链(Service Function Chain,SFC)。本文主要研究工作如下:(1)针对数据中心中基于能量感知的服务功能体部署问题,我们提出了一个新的服务功能体部署算法Merge-RD。首先针对该问题我们构建了相应的系统模型,并给出服务器和交换机设备的功耗模型,然后将基于能量感知的服务功能体部署问题建成一个0-1整数规划。此外我们定义了一个新的指标"关联度(RelationDegree,RD)"来衡量服务功能体间的流量强度和关系。然后我们提出的能耗感知算法Merge-RD将部署分成了两部分,包括服务功能体映射和链路映射。最后仿真结果显示算法Merge-RD在能耗和延时方面优于一些现有的部署算法,其相应的包交付率也有一定保证。(2)针对数据中心中服务功能体调度和服务功能链路由问题,我们使用启发式算法来解决。首先针对该问题我们构建了相应的系统模型,并给出功能体实例调度模型和服务功能链路由模型,然后将服务功能体调度和服务功能链路由问题建成一个优化问题。我们采用量子遗传算法(QGA)来同时执行服务功能体调度和服务功能链路由。最后,仿真结果显示我们的工作获得了较好的包交付率和延时分布,以及较为满意的负载分布。
[Abstract]:Today's data center deployment server and a large number of network devices, to provide a variety of network services (firewall, content caching, Wan accelerators) connection and full network. Most of the traffic for multiple service functions by processing the data center, so as to form the traditional network service chain. The traditional network service chain there are some disadvantages such as the coupling between the network device and packet forwarding efficiency, network equipment cannot be pooled and poor scalability. Network functions Virtualization (Network Function Virtualization, NFV) the traditional hardware equipment has the network function can be in the form of software deployed on standard server, service function body (Service Function, SF) can create dynamic, flexible and extensible software defined network (Software Defined; Networking, SDN) through centralized control, the control The controller can automatically generate form flow business function chain. The function of SDN and NFV chain called service chain based on function (Service Function Chain, SFC). The main works are as follows: (1) according to the service function of body energy aware deployment problem based on data center, we propose a new service function deployment algorithm Merge-RD. to solve the problem first, we construct the corresponding system model, and gives the model of power switch device and the server, and then the service function of energy aware deployment body built a 0-1 integer programming based on. In addition, we define a new index of correlation (RelationDegree, RD) "the measure of service function between the flow intensity and relation. Then we propose Merge-RD energy aware algorithm will be deployed into two parts, including the service function mapping and link mapping. Finally, simulation The results show that the proposed algorithm outperforms some existing Merge-RD deployment algorithm in energy consumption and delay, the packet delivery rate is guaranteed. (2) according to the data center service functions and service functions of the chain scheduling routing problem, we use heuristic algorithm to solve the problem. Firstly, we construct the corresponding system model. And gives the function instance scheduling model and service function chain routing model, and then the service function and service function of scheduling chain routing problem into a optimization problem. We use the quantum genetic algorithm (QGA) to perform the service function and service function of routing scheduling chain at the same time. Finally, the simulation results show that our work is obtained better packet delivery rate and delay distribution, and satisfactory load distribution.

【学位授予单位】：中国科学技术大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP308;TP393.0

【相似文献】