Algorithms for Fast Resilience Analysis in IP Networks

When failures occur in IP networks, the traffic is rerouted over the next shortest paths and potentially causes overload on the respective links. This leads to congestion on links and to end-to-end service degradation. These can be anticipated by evaluating the bandwidth requirements of the traffic on the links after rerouting for a set of relevant failure scenarios $\mathcal{S}$. As this set can be large in practice, a fast evaluation of the bandwidth requirements is needed. In this work, we propose several optimized algorithms for that objective together with an experimental assessment of their computation time. In particular, we take advantage of the incremental shortest path first (iSPF) algorithm to reduce the computation time.