Dynamically generating range of Clustered Controllers using Kernel Density Estimation for Load Balancing in a Multi-Controller Software Defined Networking

Abstract

Software Defined Networking (SDN) is one of the prominently used network paradigm. The architecture is logically separated into two planes, Data and Control plane. The forwarding devices (switches) and hosts (systems) rest in the data plane, and the centralized controller is part of the control plane. The controller is used to direct the network and modify the flow table entries in the switches. The controller and data plane nodes are connected using south-bound interface. And they communicate using OpenFlow protocol. There are various technical challenges associated with SDN. Load Balancing in the control plane is one of the research interests and we target to address the same. This paper proposes an algorithm that can dynamically group the controllers based on their loads and provide a range of possibility to apply migration algorithms to the clusters. This also provides the flexibility for load balancing and dynamic threshold.

Keywords: Controller Plane, Data Plane, Distributed, Dynamic threshold, Load Balancing, OpenFlow protocol.