page 1  (12 pages)
2to next section

Internet Architecture Workshop: Future of the Internet SystemA rchitecture and TCP/IP ProtocolsDavid L. Mills, INARC Chair; Paul Schragger and Michael Davis, EditorsUniversity of DelawareNewark, Delaware1-2 June 1989 The Internet Architecture Task Force (INARC) is a unit of the Internet Activities Board (IAB), whichoversees the research, development and standards of the Internet community and TCP/IP protocolsuite. This is a report on the INARC workshop held at the University of Delaware on 1-2 June 1989at which the future of the Internet and its protocols, as well as their relevance to other protocolcommunities, was explored and debated. IntroductionThis is a report on a workshop convened to discuss anddebate the future of the Internet architecture and TCP/IPprotocols. At the direction of the Internet ActivitiesB oard (IAB), the Internet Architecture Task Force( INARC) tracks significant happenings, research ac- tivities and evolutionary forces shaping the Internet ar-chitecture and its protocols. This workshop concentratedon the architectural and protocol issues that the Internetmust face in the years to come. The lessons learnedshould serve as guidance for research planning, policyformulation and system engineering for the Internet com-munity of the future.The first day of the two-day workshop included severalinvited or volunteered formal sessions which exemplifycurrent technology and policy issues. Due to the largenumber of volunteered informal presentations, the firstday concluded with a number of fifteen-minute mini-ses-sions. The second day included four panel discussions inwhich panelists presented some aspect of the topic andother panelists and members of the audience discussed itat length.In this report, which is condensed from the original45-page edited transcript, the words are the editors',unless noted to the contrary. While an effort was madeto balance the emphasis on all presentations, some ofthem were more provocative than others and generatedmuch heat and some light. In the concise summaries tofollow the editors tried to capture the full intent of thespeakers; but, sometimes, using words that never actual-ly came out of their mouths.Day 1: 1 June 1989SESSION 1Navigation Aids for the Future InternetDavid Mills, INARC Chair, U. DelawarePresenter's summary: In order to set the tone for thisworkshop we need to mark regions on the map which areonly beginning to be explored by Internet navigators.T his presentation raises several issues designed toprovoke interest, stimulate discussion and foster debatein the remaining sessions and panels. The following listof objectives are presented as guidelines for debate.1.Conventional wisdom cites visualization, remotesensing and national filestores as drivers for hugeand fast. The machines which require such speedsare the supercomputers and earth stations of todayand the workstations and space stations of the future.Are the Internet architecture and protocols suitablefor use on very high-speed networks operating in the1000-Mbps range and up? If the network-level ortransport-level protocols are not usable directly, canthey be modified or new ones developed to operateeffectively at these speeds?2.We occasionally see cases of Internet routing bob-bles, meltdowns and black holes, even with only 700nets and uncoordinated backdoor paths which invitesinister routing loops. Are the Internet addressingand routing algorithms adequate for very large net-works with millions of subscribers? If not, is itpo ssible to extend the addressing scope and/ord evelop new routing paradigms without startingover from scratch?3.Can the Internet model of stateless networks andstateful hosts be evolved to include sophisticatedalgorithms for flow management, congestion con-trol and effective use of multiple, prioritized paths?Can this be done without abandoning the estimated60,000 hosts and 700 networks now gatewayed int he system? Should we evolve to more statefuldesigns in order to embrace new principles of flowand policy management?4.The routing technology of the future will be coupledless to available hardware and routing algorithmsthan to issues of resource utilization, cost recoveryand administrated access. Can the existing Internet