Towards A New Internet Routing Architecture: Arguments for Separating Edges from Transit

In recent years, the size and dynamics of the global rout-ing table have increased rapidly along with an increase in the number of edge networks. The relation between edge network quantity and routing table size/dynamics reveals a major limitation in the current architecture; there is a conflict between provider-based address aggregation and edge networks ’ need for multihoming. Two basic direc-tions towards resolving this conflict have surfaced in the networking community. The first direction, which we dub separation, calls for separating edge networks from the transit core, and engineering a control and manage-ment layer in between. The other direction, which we dub elimination, calls for edge networks to adopt multi-ple provider-assigned addresses to enable provider-based address aggregation. In this paper, we argue that separa-tion is a more promising approach to scaling the global routing system than elimination, and can potentially be leveraged to bring other architectural improvements to today’s Internet that an elimination approach cannot. 1