Transport Area Directors o Allyn Romanow: allyn@eng.sun.com o Scott Bradner: sob@harvard.edu Meetings of nine Transport Area working groups and three BOFs were held during the IETF meeting in Munich, Germany. avt - Audio/Visual Transport WG The AVT working group had not planned to meet in Munich during a period of dormancy while the testing of RTCP scaling mechanisms was underway. However, one AVT session was scheduled to discuss a new problem with the RTCP scaling, plus open issues for several new payload formats that were submitted since the last meeting. The problem is a relatively minor one: if many participants leave a session at once, other participants that remain in the session may be falsely timed out. An extension to the "timer reconsideration" algorithm avoids this timeout. The new payload formats are for H.263+, MPEG4, BT-656 and QuickTime video, and DTMF tone signaling in audio. In addition, methods for repair of packet losses were discussed, including the question of how much error correction is appropriate. The meeting concluded with an update on the RTP MIB and plans to implement it. intserv - Integrated Services WG The intserv WG held a meeting at the Munich IETF dedicated to the topic of "differential services". The purposes of the meeting were to acquaint IETF participants with current technical work in this area, to consider which aspects of this technology require IETF standardization, and to discuss what the IETF should do at this time. The meeting heard presentations from Dave Clark, Van Jacobson, Kalevi Kilkki, Fred Baker, and Ed Elleson, followed by a half hour of active discussion. As an initial follow-on to the meeting, an internet draft describing the similarities and differences of the approaches presented will be prepared by the chair ipc - Internet Policy Control BOF The BOF discussed a charter for an IPC working group and came to some conclusions including that the WG should focus on RSVP policy control and not policies themselves. Jim Boyle, MCI and David Durham, Intel presented a proposal for Client/Server policy protocol (PEPCI), as a minimal functionality simpler alternative to the OOPS protocol. Shai Herzog, IPHighway, presented the second version of OOPS (OOPS-01) attempting to balance between simplicity, flexibility and scalability. Tim O'mally agreed to be Chair of an IPC WG. ippm - IP Performance Metrics BOF The IPPM meeting was held 1PM - 3PM on Monday, August 11. A major item of business was discussing the draft charter for the proposed IPPM working group, with the effort being moved from BMWG to the Transport Area in recognition of the central role transport issues play in shaping many Internet metrics. Considerable discussion centered around the word "standard" and the degree to which it should apply to the IPPM charter and the group's products. There was also discussion of the distinctions between IPPM and other IETF efforts, such as OpStat, RTFM, and BMWG. We then reviewed near-final revisions to the Framework document, prior to submitting it as an informational RFC. This was followed by discussions of operational experience with measuring one-way delays and packet loss, brief discussion of updates to the Bulk Transfer Capacity I-D, and two further experience talks, one on measurements of Internet telephony, and the other on measuring packet loss and RTT using UDP echo packets. issl - Integrated Services over Specific Link Layers WG The ISSLL WG held two meetings at the Munich IETF. The primary purpose of these meetings was to hold a full-group review a number of documents prepared by the individual technology subgroups, before holding a WG last-call on the mailing list. There are approximately 10 documents from the IS802, ISATM, and ISSLOW subgroups nearing completion. The second purpose of the meeting was to present some new work items proposed since the last IETF, and see if the group wanted to proceed with them. These items are ISATM extensions for "shortcuts" (NHRP); ISATM extensions for server-based multicast (MARS/MCS), and the creation of a new subgroup to consider IS over cable modem technology. Finally, Raj Yavatkar gave a summary of integrated services and RSVP to the IP1394 (IP over IEEE 1394 link layers) WG. Members of the IP1394 WG and the ISSLL chairs discussed forming a IS1394 subgroup in the timeframe of the Washington IETF. mmusic - Multiparty multimedia Session Control WG The MMUSIC WG had two meetings at the Munich IETF. In the first meeting Carsten Bormann briefly discussed the Internet Conferencing Architecture the revised internet draft (draft-ietf-mmusic-confarch-00.txt). An archutecture for multicast address address allocation was then presented by Mark Handley. This architecture is a three level model, with MDHCP being used to obtain an address from a local server, the servers communicate using AAP (Address Allocation Protocol), a proposed variant of SAP, and allocate individual addresses from a range which is given to them by MASC, a prefix allocation mechanism closely tied the BGMP (the Border Gateway Multicast Protocol) beihng discussed in the IDMR WG. Baiju Patel presented MDHCP and Deborah Estrin briefly presented MASC. All these three protocols are strawmen at this time. Mark Handley then presented changes to the SIP specification since the Memphis IETF. In the second meeting, Joerg Ott presented an update on the ITU-T work that is related to MMUSIC. Much of the session was taken up with Rob Lanphier's discussion of the open issues related to RTSP, and good progress was made. nat - Network Address Translator BOF Yakov Rekhter made a presentation defining NATs as a means of interconnecting disparate routing realms. Multiple routing realms are a reality and as such the IETF should address the reality. Specifically, IPSEC should support communications spanning multiple routing realms. In response, Steve Bellovin felt that re-engineering IPSEC to cope with NAT could mean lessening the overall security level between end nodes. Steve Deering presented slides pointing out that NAT can be pursued as a a transition tool prior to deployment of IPv6, but pursuing NAT as a new addressing architecture and an end-goal in itself is a bad idea and significantly inferior to the current routing realm. Pyda Srisuresh presented the internet draft that extends RFC1631 to include Network Address Port Translation (NAPT), which translates a pool of private addresses into a single address for TCP/UDP applications. Suresh listed requirements for NAT friendly applications and some of the security issues NATs have to deal with for TCP and UDP applications. In the end, Scott Bradner, the area director, addressed the audience about the rationale behind conducting the BOF and took a straw poll to assess the interest in forming a NAT work group. There was overwhelming interest to form a work group, in spite of reservations expressed by some. pint - PSTN and Internet Interfaces WG The newly-created PSTN/Internet Interworking (PINT) WG meeting took place from 15:30 to 17:30 on Wednesday, August 13. The meeting was chaired by Igor Faynberg (Bell Labs/Lucent Technologies). The WG roster registers 131 attendees. The meeting took place according to its agenda, which included a discussion of the PINT Charter, a small tutorial on PSTN and IN, two presentations on the PINT benchmark services and protocol, an introduction to the Session Initiation Protocol (SIP), and a discussion of the workplan. The major outcomes of the meeting were: 1) agreement on the urgent need for security studies; 2) agreement to study SIP on the subject of re-using it in PINT; and 3) the priority of identifying the PINT services before the protocol work starts. The initial outline of the Informational RFC was found acceptable as the starting point. rsvp - Resource Reservation Setup Protocol WG The RSVP working group met for one hour session in Munich. After a brief update on the status of the RSVP documents, currently being reviewed by the IESG, and getting working group approval for a small addition to one of the documents describing how the IANA should assign specific rsvp protocol numbers, the working group discussed its future. The result was a decision that the working group will go dormant, since its work items have mostly been accomplished or taken over by more specialized working groups. The rsvp mailing list will continue to be active. rtfm - Realtime Traffic Flow Measurement WG The WG reviewed its new charter. Presentations were given on the current meter implementations, and on three interesting research efforts. The new Meter MIB draft will go to WG last call before being submitted to the IESG as a Proposed Standard. The Architecture RFC needs small changes to bring it into line with the new Meter MIB; this will be progressed as quickly as possible. Changes to the New Attributes draft were discussed, and several interesting extensions proposed. We aim to produce a final draft of the New Attributes document for the December meeting. tcpimpl - TCP Implementation WG The TCPIMPL meeting was held 7:30PM - 9PM on Monday, August 11. We estimate the audience at around 50-75 (plus MBone viewers). The meeting began with brief discussions of additions to the "known problems" I-D, which included a description of some problems with keep-alives, and a change in the use of "Significance" when describing an implementation problem. One of the authors of a new I-D cataloging some TCP testing tools gave an overview of the I-D, eliciting discussion about some possible additions. We then engaged in lengthy discussion of TCPIMPL undertaking the task to revise RFC 2001, both to clarify (make less rigid) some of its wording, and to alter the initial window used during slow-start. Attendees debated the merits and drawbacks to a number of different ways for altering congestion control. We finished with the usual plea for volunteers to document other known TCP problems, and netted a few in the process. tcpsat - TCP over Satellite WG The TCPSAT WG met on Monday 1530-1730 in Munich. Scheduled topics included a review of working group status by Aaron Falk (TRW, WG Chair) and a briefing of the technical issues to be addressed in the working group's I-D (Mark Allman, NASA) followed by open discussion. The draft scheduled to be released in July has not come out yet. To address this, Mark Allman (NASA) has been added as a co-editor of the draft. Additionally, an interim meeting of the working group will be held in October to help advance the draft to maintain the schedule. The technical presentation was prepared by Eric Travis (doc editor) and included a summary of topics raised on the mail list as well as other issues discovered by NASA. An interesting discussion occurred during this presentation where Van Jacobson, Craig Partridge, and Tim Shepard discussed the concept of increasing the time between TCP acknowledgments as a method of improving performance under certain conditions.