基于应用层策略转换一致性的SDN流表更新方法

发布时间：2018-04-17 11:27

本文选题：软件定义网络 + 策略转换　；参考：《南京邮电大学》2015年硕士论文

【摘要】：随着互联网的不断发展,日益复杂的网络应用和管理要求促使我们寻求一种更加高效的网络体系架构,软件定义网络(Software Defined Network,SDN)应运而生。软件定义网络将控制和转发分离,改变了原有的网络体系架构,用户可以从一个全局的角度对底层网络进行控制和操作,并可以按照需求将应用层策略部署到底层网络中。由于软件定义网络起源于Openflow技术,而流表(Flow Table)的设计和使用是Openflow技术的核心,SDN通过流表的及时更新来实现对灵活的应用层策略的支持,但是如何保持流表更新一致性一直是SDN的研究关键。本文选取了三种不同应用层策略部署方案进行分析,提出一种应用层策略转换方案,在理论上解决了控制器通用性问题和策略高效部署问题。本文并进一步对应用层策略转换方案进行拓展,研究不同策略转换的流表合并和更新一致性问题。针对不同策略转换的流表合并问题,设计了一种支持多应用层策略的两阶段流表合并算法,将同一策略或者不同策略转换的相似的流表进行合并,减少了流表的数量,在一定程度上解决了底层交换机存储空间不足的问题;针对不同策略转换的流表更新一致性问题,提出了一种支持多应用层策略的流表更新一致性方案,将合并之后的流表按照各自转换的策略在不同时间段进行更新,并使数据包在其网络传输生命周期中只按照一种策略转换的新流表或者旧流表进行处理,从而保证不同策略转换的流表更新一致性,最终实现整体的应用层策略转换一致性。仿真实验证明,支持多应用层策略的两阶段流表合并算法适用于三种实验场景:第一阶段流表合并场景、第二阶段流表合并场景和两阶段流表合并场景,并在一定程度上解决了不同策略转换的流表合并问题;支持多应用层策略的流表更新一致性方案适用于两个实验场景:变更转发路径场景和变更转发端口场景,并在一定程度上解决了不同策略转换的流表更新一致性问题。
[Abstract]:With the continuous development of the Internet, the increasingly complex network application and management requirements urge us to seek a more efficient network architecture, and the Software Defined Network (SDN) emerges as the times require.The software defines the network to separate the control from the forwarding, changes the original network architecture, the user can control and operate the underlying network from a global perspective, and can deploy the application layer policy to the underlying network according to the requirement.Because the software definition network originates from the Openflow technology, the design and use of flow table is the core of Openflow technology to realize the support of flexible application layer policy through the timely updating of the flow table.However, how to keep the consistency of stream table update is the key of SDN research.In this paper, three different application layer policy deployment schemes are selected for analysis, and an application-layer policy conversion scheme is proposed, which solves the problem of the universality of controller and the efficient deployment of policy in theory.In this paper, the application layer policy transformation scheme is further extended to study the consistency of stream table merging and updating for different policy transformations.In order to solve the problem of stream table merging with different policy transformation, a two-stage flow table merging algorithm supporting multi-application layer policy is designed, which combines the same policy or similar stream table transformed by different policies, thus reducing the number of stream tables.To some extent, the problem of storage space shortage in the bottom switch is solved, and a stream table update consistency scheme supporting multi-layer policy is proposed to solve the consistency problem of stream table update for different policy conversion.The merged stream tables are updated at different time periods according to the policies they have converted, and the packets are processed only according to a new stream table or an old stream table converted by one policy during their network transport life cycle.In order to ensure the consistency of stream table update of different policy transformation, the overall consistency of application layer policy transformation is finally realized.Simulation results show that the two-stage flow table merging algorithm which supports multi-application layer strategy is suitable for three experimental scenarios: the first stage flow table merging scenario, the second stage flow table merging scenario and the two-stage flow table merging scenario.To some extent, the flow table merging problem of different policy transformation is solved, and the consistency scheme of stream table updating supporting multi-application layer policy is applied to two experimental scenarios: changing forwarding path scenario and changing forwarding port scenario.To some extent, the consistency problem of stream table updating with different policy transformation is solved.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP393.02

【相似文献】