Online OSPF weights optimization in IP networks

The high volatility of traffic patterns in IP networks calls for dynamic routing schemes allowing to adapt resource utilization to prevailing traffic. In this paper, we focus on the problem of link weight optimization in OSPF networks where the traffic is routed along shortest paths according to the link metrics. We propose an online approach to optimize OSPF weights, and thus the routing paths, adaptively as some changes are observed in the traffic. The approach relies on the estimation of traffic demands using SNMP link counts. Experimental results on both simulated and real traffic data show that the network congestion rate can be significantly reduced with respect to a static weight configuration.