| Software Defined Network (SDN), regarded as a newly designed network architecture, is capable of realizing flexibly scheduling the network traffic and openly programming the network behaviors through decoupling the control plane and the data plate of the network devices. However, a large quantity of the problems are remained to be solved, especially for the expansion of the control plane and the promotion of the data forwarding capability on the data plane.The thesis first introduces the state-of-the-art research on SDN including the background and the promotion of the system performance. The various performance optimization schemes of the controllers and the OpenFlow switches is then elaborated, with their advantages and disadvantages well organized in the thesis.The main part of the thesis lays on the design of an adaptive scheme for data forwarding in SDN. The design is carried out in two parts, namely the resource trade-off scheme between the controller resource and the switch flow table resource and the idle timeout reset optimization of the flow tables. The design achieves higher availability and the matching ratio of the flow table resource as well as alleviates the congestion of the control channels.Moreover, the thesis also analyzes the application of the multi-stage flow table technology, such as the partition method of the flow table and the capacity assignment and the sorting between each flow table, based on which the thesis then proposes specific solutions and design principles, as well as an OpenFlow-based architecture of variable multi-stage flow tables. The result indicates our method shows remarkable improvement of the data forwarding success of the multi-stage flow tables.Finally, the thesis views the prospect of the soft-defined networks. In addition, the relative research in the thesis is well summarized and the future directions are also discussed. |
