Research Of Network Forwarding Policy Management In The SDN Controller

The current networks’ weakness of enormousness, complexity and difficulty to manage and deploy conflicts sharply with the growing network scale and user needs. Software-Defined Network(SDN) proposed a new network architecture, one basic characteristics of which is the use of open and standard interfaces, such as OpenFlow, ForCES, to replace various equipment suppliers’private configuration methods, in order to simplify the network configuration tasks, which provides a new solution for the future network management.To further open network devices and enhance the flexibility, south protocols, such as OpenFlow, ForCES, have defined large number of complex and low-level interfaces, which brings many challenges to SDN programmer. Therefore, how to simplify the programming in SDN has become a hotspot currently.Aiming at this problem, this thesis firstly analyzes and compares the current SDN-oriented programming languages, such as NetCore and Pyretic, and on this basis, designs a Forwarding Policy Management Subsystem for OpenDaylight Controller, so as to enable it to provide SDN network control functions based on advanced network programming languages. Secondly, this thesis detailly describe the syntax, semantics and implementation of Forwarding Policy Service.Secondly, this thesis detailedly describes the syntax, semantics and implementation of Forwarding Policy Service, which is a module of Forwarding Policy Management Subsystem. And considering the demand for Forwarding Policy Service, this thesis modifies part of the policy combination algorithm of NetCore so that it can further supports detection and prompt of the policy conflict.And then, in order to improve the adaptability of Forwarding Policy Management Subsystem, this thesis selects ForCES protocol as the research object and conducted intensive studies of some key issues for conversion of forwarding policy.Finally, this thesis tests the policy combination algorithm based on the Pyretic Project. Experimental results show that the modified policy combination algorithm can effectively detect and prompt policy conflicts.