Software Defined Network (SDN) makes routing decisions according to the shortest path. This routing scheme does not consider link information such as link residual bandwidth, forwarding delay, packet loss rate and so on. In the case of large data flow, it is easy to cause network congestion, thus reducing the network quality of service assurance. For this reason, using the ability of SDN network controller to monitor the underlying network information and control the route forwarding, we try to set multiple objective functions according to different QoS parameters, and use the non-dominated sorting multi-objective optimization genetic algorithm (NSGA II) based on the elite strategy to make routing decisions. Compared with other routing algorithms, the proposed multi-objective optimization genetic algorithm reduces forwarding delay and packet loss rate, so can effectively prevent network congestion.

Aiming at the common load balancing problems in the network, a multi-path load balancing strategy (MPF-MLBS) based on multi-performance factor (MPF) for Software Defined Network (SDN) networks is proposed, which can be divided into two stages: algorithm design and strategy implementation. In the algorithm design stage, the advantages and disadvantages of the existing load balancing algorithms are analyzed, combined with the characteristics of the SDN network architecture, and the bandwidth, delay and link rate of the network link are comprehensively considered. Based on this, multiple performance factors are defined, and a load balancing algorithm based on multiple performance factors (MPF-CMP) is designed and implemented. In the strategy implementation stage, build a multi-path network topology based on the SDN architecture, and use the depth-first traversal algorithm to traverse the global network to obtain the required link information; Subsequently, the MPF-CMP algorithm and OpenFlow group table technology are combined to complete the proportional distribution of network traffic to each available path, thereby achieving multi-path load balancing of the SDN network. The simulation experiment results show that the strategy effectively exerts the SDN controller's overall network monitoring and scheduling functions, can obtain link information in real time and distribute and transmit network traffic according to the situation. The overload of a single path is avoided, the data packet transmission volume of all available paths is effectively increased, the flow transmission efficiency of the entire network is improved, and the multi-path load balancing of the SDN network is realized.