Research About The Controller Placement Problem Of Availability-Based SDN

SDN (Software-Defined Network) as a new network architecture, is to build a new, stable, reliable and scalable network. It can not only improve the efficiency of network equipment operation and maintenance, liberating IT staffs, but also provide dynamic intelligent network service-level applications. The forwarding device and the controlling device are separated in it, such distributed network architecture reduce the burden on the control-layer and make flexible control on the network traffic. With the continuous expansion of the network size, single-point controller unable to meet the needs of the control platform apparently, once a single point of failure occure, it will cause the entire network paralysis. Considering the high-availability of network system, this paper take the information propagation delay between controllers in WAN as the key factors that affect the performance of whole control network system in SDN, and make an in-depth research on the controller placement deployment problem.This paper analyzed the impact of the controller placement problem on the control-layer performance, and put forward a reasonable controller deployment strategy in SDN network. On the basis of this issue, considering the high-availability requirements of the network system and using the redundancy backup technology, this paper proposed a novel controller redundant strategy "ORDS" and an optimal backup deployment algorithm "OBA", so as to greatly reduce the information interaction distance between controllers and shorten the control information interaction time, and then make the network system can be more efficient and reliable to meet the needs of internet users. Finally, this paper tested the function and performance of the controller redundant strategy "ORDS" and the optimal backup deployment algorithm "OBA", fully verified the feasibility and correctness of the above key technologies. At the same time, this paper made a prospect for future research work based on the problems which still exit in the test.