A Scalable Routing System Design for Future Internet

Internet routing is at an important crossroad. The current global routing table, which is largely based on IPv4 addresses, has been growing at an alarming rate over the last few years, despite the constraints by the shortage of IPv4 addresses. IPv6 removes the address shortage problem, however its deployment may potentially further exacerbate the routing scalability challenges facing us today. In this paper, we first examine and describe the root causes of the routing scalability problem and then discuss a promising direction towards an effective solution. The explosive growth of the Internet over the last decade made it no longer feasible to perform global routing based on all end user IP address prefixes. Yet at the same time, we must preserve the end-to-end model of the Internet architecture. We sketch out a basic approach to an effective solution which is to separate globally routable addresses (GRA) from globally deliverable addresses (GDA). This separation of address space can simultaneously achieve the goals of improved routing scalability, ease of site-multihoming without using multiple addresses, and elimination of the need for user renumbering when changing providers. An interesting aspect of this approach is that it both facilitates the deployment of IPv6 at edge sites and also does not require immediate changes at large IPv4 deployed bases.