Abstract
This paper studied the cross-domain load balancing communication mechanism based on the game model. Firstly, the problem of routing between SDN/IP heterogeneous network inter-domain communication is analyzed. The SDN/IP cross-domain hybird network communication model is proposed. The network abstraction idea is introduced. The OpenFlow controller can be supported by the three-layer routing module and edge nodes. IP routing protocol to synchronize network reachable information between hybrid networks. In the Layer 3 routing module, the routing engine sub-module parses and responds to routing messages, and abstracts the OpenFlow network as a standard "Router". At the same time, three sub-modules of edge management, routing processing, and routing engine management are designed on the controller side. Forward control of inter-domain datagrams of SDN internal networks and management of routing engine sub-modules. Then, a feedback mechanism is added at the edge node location to actively detect and report the edge location state information to expand SDN impact. Secondly, for the problem of elephant stream forwarding in SDN/IP cross-domain network, a multi-path load balancing communication mechanism oriented to cross-domain is designed. The mechanism monitors the elephant flow in the network through the traffic monitoring module in real time, and proposes a multi-path dynamic load balancing algorithm to achieve network load balancing. Firstly, according to the real-time QoS index of the link, multiple minimum cost paths are calculated, and the idea of stream splitting is introduced. Multiple paths are used to bear an elephant stream, which reduces the burden of a single path and improves the utilization of network resources. Then the game model based auction model is introduced to calculate the offload strategy to optimize the maximum link utilization of the network. The simulation results show that the algorithm can effectively balance network traffic, meet the QoS requirements of the service, and increase the network resource utilization rate to 83%.