Abstract
The Border Gateway Protocol (BGP) is the primary inter-domain routing protocol used in today's global Internet. Despite many refinements over the years, studies have shown that BGP control message traffic in the Internet is much higher than one would expect. This has prompted further research to attempt to explain this surplus of traffic with the hope that it can be corrected. Though many hypotheses have been suggested, the size and complexity of the Internet have made it extremely difficult to sort out just what's going on. Current research typically focuses on studying routing data exchanged between service providers. However, advances in the field of parallel discrete event simulation have opened the door to the possibility of simulating very large networks (containing millions of nodes). Simulating BGP on such a network would allow testing and observation of the protocol's behaviors under controllable conditions, improving the chances of discovering the causes of the extra traffic. In this talk, I will describe BGP, including its purpose, expected behavior, and associated mysteries; introduce the Dartmouth Scalable Simulation Framework (DaSSF); and discuss how DaSSF can be used to study BGP's behavior.