Towards Reproducing Inter-Domain AS Paths

Today’s Internet is composed of approximately 20,000 competing autonomous systems (ASes) which have to cooperate with each other to provide global reachability. Each AS is controlled by different administrative entities, and therefore encodes various economic, business, and performance decisions in its routing policy. Unfortunately, the complex interactions arising from such a distributed system are still poorly understood. In this thesis, we present a methodology to reproduce the BGP inter-domain routing decisions as realistically as possible by means of simulation. We start with a simple simulation model, relying solely on an AS graph which is derived from BGP table dumps. Due to missing policies, the routes chosen in the simulation will be different from those observed in reality. In an iterative process, we try to refine our simulation to accommodate the simulated routing decisions with those observed in reality. Such an Internet-wide simulation technique can be very valuable for ISPs to diagnose problems, optimize network performance, identify peering partners and provide better connectivity to the rest of the Internet. In addition, researchers can benefit from such a model. The design of more sophisticated traffic engineering techniques, as well as a better understanding of the interplay between the routing layer and overlay networks are only two applications.