Fuzzy Logic Load-Balancing Strategy Based on Software-Defined Networking

Traditional load balancing hardware is expensive and lacks scalability and flexibility. We propose a load balancing strategy based on fuzzy logic (LBSFL), which exploits the control and forwarding separation architecture characteristics of software-defined networking (SDN). First, the fuzzy membership function that affects the performance parameters of the server load is analyzed. Based on this, the load state of the virtual server is evaluated through fuzzy logic. Then the centralized control capability of SDN's controllers for the whole network is utilized to monitor virtual server information in real time and to schedule virtual server tasks. Individual servers can be hibernated or restarted, to save power or to increase performance as necessary. Finally, the dynamic balance between the overall load, performance and energy consumption is realized. Simulation experiments showed that the proposed strategy improves overall performance of the network, especially when dealing with communication-intensive tasks and using a high-latency network.