Queuing-aware Routing Algorithm in Software Defined Networks

Abstract

Routing Algorithm is an important mechanism which decides data transferring path in a network. Several existed routing algorithms are derived from a fundamental of graph theory with shortest path approach. A lot of additional network metrics were applied to serve the best quality of service (QoS) to end devices. However, a selected path from the existed algorithms can suffer from additional end-to-end latency for long time periods when the number of traffic flows increases. This situation is called the bufferbloat problem which is caused by excessively large queue in buffer. In this paper, we proposed a novel queuing-aware routing algorithm in software defined networks. The proposed algorithm leverages a capability of centralized system to gather directly the buffering information in real-time to calculate the shortest path with acceptable buffering occupancy level. The simulation result shows that the proposed algorithm outperforms the traditional shortest path approach in term of overall throughput significantly.