new addressing and routing design called the Less-Is-More Architecture (LIMA) is proposed as an inter-domain solution for a future Internet. Unlike recently proposed identifier-locator split solutions, LIMA uses just (topological) location-independent names and location-dependent addresses. The feasibility of using a policy combination of restricting stubs to provider-aggregatable addressing only, and disallowing stub-level reachability from being propagated into the global routing tables, is studied. This policy combination results in significantly smaller global routing tables but creates four challenges of address renum-bering (when stubs change providers), multihoming, mobility, and traffic engineering. Solutions to these challenges include the use of multiaddressing, name based sockets, a LIMA concept of address dismemberment, transport protocols such as SCTP that are capable of dynamic address reconfiguration, and new management-plane and control-plane procedures. Preliminary RIB data analysis quantify the benefit of LIMA in global routing table size reduction (to 6815 entries from today's 335K entries), and a cost of LIMA in terms of number of provider changes made by stubs in the last six months (about 2450 provider changes per month across 33K stubs).
The different versions of the original document can be found in: