TCP and UDP with Bigger Addresses (TUBA), A Simple Proposal for Internet Addressing and Routing

摘要

The Internet is approaching a situation in which the current IPaddress space is no longer adequate for global addressingand routing. This is causing problems including: (i) Internetbackbones and regionals are suffering from the need to maintainlarge amounts of routing information which is growing rapidly insize (approximately doubling each year); (ii) The Internet isrunning out of IP network numbers to assign. There is an urgentneed to develop and deploy an approach to addressing and routingwhich solves these problems and allows scaling to several ordersof magnitude larger than the existing Internet. However, it isnecessary for any change to be deployed in an incremental manner,allowing graceful transition from the current Internet withoutdisruption of service. [1]This paper describes a simple proposal which provides a long-termsolution to Internet addressing, routing, and scaling. Thisinvolves a gradual migration from the current Internet Suite(which is based on Internet applications, running over TCP orUDP, running over IP) to an updated suite (based on the sameInternet applications, running over TCP or UDP, running over CLNP[2]). This approach is known as 'TUBA' (TCP & UDP with BiggerAddresses).This paper describes a proposal for how transition may beaccomplished. Description of the manner in which use of CLNP,NSAP addresses, and related network/Internet layer protocols(ES-IS, IS-IS, and IDRP) allow scaling to a very large ubiquitousworldwide Internet is outside of the scope of this paper.Originally, it was thought that any practical proposal needed toaddress the immediate short-term problem of routing informationexplosion (in addition to the long-term problem of scaling to aworldwide Internet). Given the current problems caused byexcessive routing information in IP backbones, this could requireolder IP-based systems to talk to other older IP-based systemsover intervening Internet backbones which did not support IP.This in turn would require either translation of IP packets intoCLNP packets and vice versa, or encapsulation of IP packetsinside CLNP packets. However, other shorter-term techniques (forexample [3]) have been proposed which will allow the Internet tooperate successfully for several years using the current IPaddress space. This in turn allows more time for IP-to-CLNPmigration, which in turn allows for a much simpler migrationtechnique.