Date: Sun, 8 Jan 95 21:35 PST To: postel@ISI.EDU, iab@ISI.EDU From: AWeinrib@ibeam.jf.intel.com (Abel Weinrib) Subject: Minutes for Dec. 9 IAB Meeting (FINAL) ============ MINUTES FOR DECEMBER 9, 1994 IAB MEETING PRESENT: Brian Carpenter Steve Crocker Robert Elz Elise Gerich Christian Huitema Jon Postel Yakov Rekhter John Romkey Dave Sincoskie Mike St. Johns Abel Weinrib Lixia Zhang NEXT MEETING: Teleconference 10-12 EST Tuesday January 10, 1995. NEW ACTION ITEMS: * Christian Huitema: Set up an organizing committee for an IAB workshop on measurements of Internet service. * Christian Huitema: Prepare letter to the Internet Society stating that the IAB believes that ISOC should be careful not to "sell its soul" and asking how it plans to support the IETF secretariat. *Christian Huitema: Prepare survey of how to use existing router MIBs to collect accounting information. *All: Provide feedback to Yakov on his paper about "routing architecture for a multi-provider, international internet." *Jon Postel: Prepare a draft of a statement for ISOC to sign re. proper use of its ownership of IPv6 addresses. OLD ACTION ITEMS: * Dave Sincoskie: Call Tony Rutkowski to offer him ownership of the Internet usage survey. * Christian Huitema, Jon Postel and Steve Crocker: Write statement re. "branding" RFCs with a protocol number and creation of a web page to support this. * Christian Huitema: Get final text of liaison document from Stev Knowles and publish it. * Lixia Zhang: Review DOD versions of the IETF standards documents. * Dave Sincoskie: Follow up on standardizing S/key within the IETF. * Jon Postel (IANA): Report on problems with the current DNS registry process and possible solutions. * Lixia Zhang, Yakov Rekhter and Phill Gross: Write discussion paper on the impact of commercialization on the Internet. * Christian Huitema: Write discussion paper on the integration of services and its impact on usage and models of usage. * Yakov Rekhter: Revise RFC 1560. * Christian Huitema and Steve Crocker: prepare a brief note outlining follow-up to security retreat. 1) Clear the breadcrumbs and bash the agenda. ------------------------------- =============================== 2) Our main topic: the evolution of the architecture under the pressure of commercialization. We will have had 3 introductory talks during the open plenary, hopefully some response and input from int-serv... ------------------------------- The first topic discussed was the problem of defining what it means to provide Internet service. Currently, there is no standard way to talk about the service that a service provider offers, and no standard measurement tools to measure the performance received. The existence of such specifications and tools would let users and providers agree upon what service is to be offered, and would help to identify what is the problem when connection to the Internet is poor. The basic problem is to define performance measures that correlate with "good performance" for various uses of the Internet. The outcome is likely to be a description of the load and measurements to be taken, perhaps involving instrumented TCP, telnet and WWW sessions. There will also probably be different ones for end-users and between carriers. Given the specification, different people can then build measurement tools. A number of members of the Internet research community are interested in this problem. Sally Floyd at LBL is looking into it, and Matt Mathis (of Pittsburgh Supercomputer Center) is also interested if he can find the time. It was decided to try to organize an IAB workshop on this topic. It would bring together experts in testing, queueing theorists, service providers and users. There is an operators conference in February, and it would be desirable to have something to input into that community by that time. However, that appears too soon, and the current goal is to hold the workshop in early March, at least a month before the next IETF meeting to allow the results of the workshop to potentially feed into activities at the meeting. Exact dates and location are still open. Christian will set up an organizing group for the workshop. .......... The next issue addressed was the question of who "owns" the root of the IPv6 address tree. It was generally agreed that the IANA administers it, but some entity need to have the authority to set the rules by which addresses are allocated. That entity should be the Internet Society with the guidance of the IAB, but the Society should not use its ownership of the addresses as a source of income or other improper influence. For instance, there should be no cross subsidies, and no monopolistic or predatory pricing policies should be allowed to develop from control of the address assignment. The best approach appears to be for the Society to use separate accounting for whatever (presumably small) revenue it may get from administering the addresses. The IAB will prepare a draft statement to this effect for the Society's signature. Other issues that were raised, but not fully resolved, include: Are addresses leased or owned by the user? Is it possible to take them back when they are no longer being used? It was generally agreed that the highest level assignments to registries should be relatively permanent, but there still remains the issue of what happens if a registry goes out of business or is not using a large block. .......... The next topic of discussion was provider vs. client owned addresses. This topic had evoked significant discussion during the open IAB meeting, with many members of the audience expressing dismay at the fact that the proposed (provider-based) IPv6 address assignment scheme ties people to providers. While renumbering upon changing providers will be possible once automatic renumbering tools are available, it makes users critically dependent on tools that will be used only rarely, and makes it difficult to have multiple providers. (It was pointed out that provider addressing is only one of potentially many schemes, but so far no other scheme has been written down.) Trying to ameliorate the concerns by introducing addresses with no topological significance leads to the question of how large a flat address space can be routed? With geographic addressing, there is a NAP in each metropolitan area that hold a monopoly--thus perhaps leading to requirements for "neutral" interconnection at the NAPs. At this point, the discussion on address allocation schemes was tabled because there are working groups studying this problem and few IAB members had yet read the relevant documents. .......... The final issue was whether there is a need for the IETF to be working on accounting tools for the Internet. The basic issue is that providers may want efficient accounting so that they can apply whatever billing policies they see fit. However, it is not obvious that this is a technical issue that needs to be addressed by the IETF. In addition, much of the information required for accounting appears to already exist in MIBs already existing on routers (packet counts, etc.). =============================== 3) Review of open action items: ------------------------------- =============================== 4) Clean up a few administrative issues: ------------------------------- Steve Crocker will serve as the IAB representative to the Nominations Committee. =============================== 5) Other issues ------------------------------- Copyrights and Domain Name Service names. This remains a problem. Right now, the registry is putting a disclaimer on the registration form saying that you should not register names that you don't have rights to. It was pointed out that a very large number of .com domains have been registered in the last year, and that almost all of them have been done correctly. The problem is that a couple of the mistakes have become very visible. A related problem is how to make the .com domain less flat so that there is less competition for names. One possibility is to create subdomains for different industries, but this leads to considerable ambiguity. The IANA would like suggestions on how to create more top-level domains. In addition, some form of registration fee would help support the Internic, thus potentially dealing with the backlog and reducing the chance of error. However, in this case it was generally felt that there then needs to exist alternative registries so that no single entity has a monopoly. The discussion then turned to ways to financially support the IETF secretariat and the related issue of ISOC financing. The ISOC does not have the means to fund the entire IETF operation, and has other higher priorities for its current limited funds. Fortunately, US government funds were adeqate for 1994 needs. However, it was generally felt that ISOC funding of the IETF secretariat is the right model in the long term. In the near term, it was agreed that the IETF registration fee could be raised to more accurately reflect the actual cost of organizing the IETF meetings; furthermore, US government support is likely to continue for some time and there is some chance that the European community could be persuaded to provide some funding as well. Finally, it was reported that the IEEE Communication Society is forming a new "Internet Technical Committee", with a charter to stimulate technical exchanges between the Internet and telco communities. =============================== =============================== These minutes were prepared by Abel Weinrib, AWeinrib@ibeam.intel.com. An on-line copy of these and other minutes are available in the directory ftp.isi.edu:pub/IAB.