Load Balancing Based Intelligent Software Defined Networking (SDN) Controller

Abstract

The data plane and the control plane are divided by the Software Defined Networking (SDN) paradigm, where an SDN controller governs the operation of the switches it controls while also receiving demand from associated switches. Switches are dynamically controlled by their respective reassigned to be able to distribute the volume of work among controls for SDN. The majority of creative solutions for assignments require a crucial component of collect data regarding switches that need to be reassigned in order to execute load balancing. An individual super controller is utilized data from all switches and gathers information for all controllers, there is a scalability issue that arises as the number of controller’s increases. When assigning switches to controllers in a big network, distances between them can occasionally be a problem.
To solve the above problem, this study proposed an intelligent clustering network based on k-means clustering. Under the OpenFlow protocol, Forwarding Table Management Standard Flow Table, and Flow Management, the advantages demonstrate the intelligent network of control concentration. Therefore, the goals were to balance the load by adjusting the cluster node load method and comparing the performance with the traditional controller cluster and single controller. The experimental results reveal that the k-means clustering method proposed in this study can effectively reduce the packet loss rate or delay. And theoretical research shows that our approach delivers a close to ideal result. According to simulation results, our dynamic grouping enhances stable clustering by scaling factor of five, which is significant.