Scaling SDN network with self-adjusting architecture

Abstract

In Software defined Networking (SDN) architecture, the control and data planes are decoupled, which makes controller manage the data plane in a logically centralized way. However, there is a problem in scaling SDN because of the centralized control plane. In this paper, we proposed a self-adjusting architecture based on Pairing heap to scale SDN network. This idea originates from network virtualization function(NVF). We see the whole network as a huge heap and divide it into several sub heaps repeatedly until get the basic units of physical switches in the network. We also applied an optimized Dijkstra's algorithm based on Pairing heap. The algorithm considers the node weight for a graph derived from SDN topology. We also compare it with original Dijkstras algorithm and results indicate the optimized Dijkstras algorithm outperforms the original one when the network is dense.