Research On SDN Openflow Controller Multi-granularity Matching Rules

With the high-speed network technology development, a lot of rich and colorful applications come into being. Traditional network architecture is less able to meet the demand of the rapid growth of business, especially cloud computing, the dynamic management of network resources optimization allocation etc. The novel technology such as SDN (Software Defined Network) and NV (Network Virtualization) can meet these requirements. SDN decouples the data plane and control plane, provides control plane user-oriented programmability, according to customer’s diversified needs enough to provide a variety of applications. NV abstracts the underlying equipments, resources, physical resources into logic resources, so as to get flow isolation, improve efficiency and management flexibility, and reduce the cost, which features is used to implement multi-tenancy isolation between the business needs.Based on SDN and NV, the main content in this thesis is aimed to use SDN south interface OpenFlow protocol provides multi-granularity matching rules to implement multiple granularity strategies, including network traffic isolation in multiple virtual networks and centralized control avoid conflict strategy. Using Floodlight as controller, Mininet builds up network simulation topology, FlowVisor realizes the establishment of a slice network to validate the multi-user multiple granularity strategy implementation is effective in using the ideal of SDN and NV.This thesis firstly introduces the research background and current research at home and abroad. Secondly, this thesis introduces framework of SDN OpenFlow protocol and controller Floodlight evolving from v1.0 to v1.4. Then, thesis introduces the basic concepts and FlowVisor tool of network virtualization. Later, with the help of the sFlow tool, based on SDN multi-granularity and NV teamwork strategy can work well. Through Mininet simulation environment, the system framework and design is introduced in detail to test control layer in the environment of network virtualization for multi-user multiple granularity strategy isolation, and the feasibility and effectiveness of the mechanism is validated. Finally, the work in this thesis are summed up, and the shortage is analyzed.Improving the work in future.