Modeling as a necessary step for understanding Internet-wide route progagation

Abstract

Most of the insight we have today about the Internet-wide route propagation comes from studies performed more than 5 years ago [1], [2], [3]. Since then, insight about the routing system has largely focused on the behavior of individual ASes [4], [5], [6], [7] or particular prefixes [8]. The way BGP routes propagate across the Internet, how fast, and whether BGP does it reasonably well are largely open questions. To argue in favor or against potential alternatives to BGP, the research community must be able to provide answers to such questions. We argue that since measurements cannot reveal the whole picture of the Internet route propagation, building models of the Internet is one of the most reasonable ways to improve our understanding of this aspect of the interdomain routing system. [9] constitutes a first step in this direction.