Towards A New Internet Routing Architecture: Arguments for Separating Edges from Transit Core
Internet routing is facing a serious scalability problem. The size and dynamics of the global routing table have increased rapidly along with an increase in the number of edge networks. A major limitation in the current architecture is that there is a conflict between provider-based address aggregation and edge networks' need for multi-homing. This calls for either separating edge networks from the transit core, and engineering a control and management layer in between. Or for edge networks to adopt multiple provider-assigned addresses to enable provider-based address aggregation. Separation is a more promising approach to scaling the global routing system than elimination. In addition to solving the primary routing scalability problem, this separation solution offers a number of other advantages: enabling end-path selection and multipath routing, raising the barrier against malicious attacks to the routing infrastructure, allowing the edges and the core to freely evolve independently from each other, and providing a boundary around the transit core in the form of a mapping service, where various new security and control functions can be easily implemented. However, the Internet has grown so large over time that it is now technically and economically infeasible to have all IP devices continue to live in the same address and routing space. Hence, a separation, along with a new mapping service, is both necessary and justified.