On Stable Route Selection for Interdomain Traffic Engineering: Models and Analysis

BGP route selection is increasingly being used by ASes to achieve interdomain traffic engineering objectives. One fundamental feature of route selection for interdomain traffic engineering is that routes for a set of destinations be chosen jointly to satisfy traffic engineering constraints and meet traffic engineering objectives. In this paper, we present a general model of route selection for interdomain traffic engineering by allowing the routing of multiple destinations to be coordinated. We identify potential routing instability and inefficiency by showing that there exist networks where the interaction of the route selection of multiple destinations can cause routing instability, even though the networks are guaranteed to converge to a unique route selection when each destination is considered alone. We derive a sufficient condition to guarantee routing convergence. We also show that the constraints on local policies imposed by business considerations in the Internet can guarantee stability without global coordination. Using realistic Internet topology, we evaluate the extent to which routing instability of interdomain traffic engineering can happen when the constraints are violated. We further generalize the preceding model under two extensions. One, we investigate the general model that the preference of an AS depends on not only its egress routes to the destinations but also its inbound traffic pattern. Second, instead of studying a specific route selection algorithm, we study a general class of route selection algorithms which we call rational route selection algorithms. We present a sufficient condition to guarantee routing convergence in a heterogeneous network where each AS runs any rational route selection algorithm. We also show that there ∗Computer Science Department, Yale University, New Haven, CT 06520-8285, USA. Email: [email protected]. Supported by NSF grant ANI-0207399. †Computer Science Department, Yale University, New Haven, CT 06520-8285, USA. Email: [email protected]. Supported by NSF grant ANI-0238038. ‡Computer Science Department, Yale University, New Haven, CT 06520-8285, USA. Email: [email protected]. Supported in part by NSF grants ANI-0207399 and ANI-0238038. §Networking Research Lab, Bell Labs, Lucent, Holmdel, NJ 07733-3030 Email: [email protected]. ¶Department of Computer Sciences, The University of Texas at Austin, Austin, TX 78712, USA. Email: [email protected]. ‖Computer Science Department, Yale University, New Haven, CT 06520-8285, USA. Email: [email protected]. Supported in part by NSF.