Despite the pressing need to secure routing, none of the existing secure variants of BGP has been widely deployed. Due to the size and decentralized nature of the Internet, it became clear that any viable secure routing protocol must offer benefits also in its early stages of deployment. In order to determine when the protocols are not adoptable, we quantify the benefits offered by a partial deployment of an Idealized Secure BGP which is able to detect malicious routes with perfect accuracy. We also quantify the benefits of an imperfect version of the protocol. Subsequently, we conclude that even the best protocols which simply detect and avoid bogus routes do not offer good security performance except in limited scenarios. We offer alternative designs, and hope that our insights will result in a new secure routing protocol that will be more attractive to early adopters.
Martin Suchara, Ioannis C. Avramopoulos, Jennifer