IETF trip report ================ The 24th IETF was held in the Hyatt Regency hotel in Boston (actually in Cambridge) from July 13th to the 18th. There were 675 attendees, an all time record. The IETF consisted of several plenary sessions and technical reports, and some 80 working group sessions and "Birds Of a Feather" (BOF) sessions, also an all time record. The conference hotel was connected to the Internet by means of a 10 Mbits/sec laser link between the hotel and MIT, a few blocks down the road. The following report consists of reports of areas of interest for RIPE, the RIPE NCC and of personal interest. Please note that the report is a personal impression. If you have any questions about all of this please send mail either to me directly <marten@ripe.net> or the NCC <ncc@ripe.net>. Shared Whois Project ==================== MERITs taken up a project to align the network databases of the US NIC, RIPE and MERIT. Their first approach is to load all databases into a temporary X.500 database and find all conflicts. The conflicts will then be solved between the three partners in this project. The initial conflicts to resolve are network names and responsible persons. The initial structure used for database alignment is a non-standard structure designed by MERIT, which will be used ONLY for the database alignment. A more definite structure will be designed in close cooperation with the OSI Directory Services group of the IETF. A longer term project is to find a means of actually distributing the databases among the (initially three) partners. The basic idea was to use X.500 for this distributed database, but it will need quite some research to find an appropriate structure. MERIT will take the lead in the whole project, the US NIC and the RIPE NCC will of course work in close relationship with them. contact persons: RIPE NCC <ncc@ripe.net> Sheri Repucci <smr@merit.edu> BGP and BGP deployment ====================== The BGP-4 specifications have been finalized at this IETF and will be send out to the IESG for proposed standard. BGP-4 will support classless addressing and routing following the CIDR proposal. The first BGP-4 field tests between several implementations are planned for January 1993. MERIT will coordinate interoperability tests. Peter Lothberg gave a presentation of the current situation of EBONE, specifically the BGP implementation in EBONE. Currently, the EBONE BGP network seems to be the biggest network running BGP-3. The US BGP network is mostly using BGP-2 and BGP-1. Gated and Cisco implementation of BGP-3 seem to be stable. references: internet-drafts: draft-ietf-bgp-bgp4-00.txt draft-ietf-bgp-bgp4-01.txt draft-ietf-bgp-experience-00.txt draft-ietf-bgp-usage-01.txt Global IP addressing ==================== It was generally accepted that CIDR will be used as the interim solution, until a new version of IP will solve the addressing and routing problems. The IP addressing plan that will be used, will be a merger of two proposals written by Daniel Karrenberg/Bernhard Stockman and Tony Li/Yakov Rekhter. Bernhard Stockman and Tony Li will merge the two documents, and hope to send the document as an informational RFC within 2 months. references: internet-drafts: draft-karrenberg-proposal-00.txt draft-rekhter-ipaddress-guide-01.txt Global Internet Exchange (GIX) ============================== There was quite some discussion on a GIX which should provide a interconnection point for everyone who whishes to connect globally. The model is similar to the NSFnet "NAP" proposal. No official sessions were held. The IEPG has taken a lead in coordinating and engineering the GIX. The GIX will most probably be located at the US east-coast. contact person: Bernhard Stockman <boss@ebone.net> IPv7 ==== The "hottest" item at this IETF was the IAB position with regards to IPv7. The IAB retracted its position on the first day of the IETF, to align with the IESG recommendation. That recommendation says that we should start CIDR now, and wait until the November IETF to make a decision on what IPv7 will be. There have been several proposals for IPv7 or parts of it (PIP, NIMROD, TUBA and others). They can all be found in the internet-drafts directory of your favorite internet-drafts server (ftp.ripe.net for instance). Following the IPv7 statement of the IAB, in the Thursday evening open plenary, there was a large debate on the relation between IAB, IETF, ISOC and other bodies. references: internet-drafts: draft-tsuchiya-pip-00.[ps|txt] draft-tsuchiya-pip-overview-01.[ps|txt] draft-callon-routing-00.ps draft-chiappa-ipaddressing-00.txt draft-chiappa-routing-00.txt draft-iab-ipversion7-00.txt rfc1347.txt Policy based Routing ==================== There was a lot of action in the field of policy based routing. It really is something that people need now, and we will certainly need even more in the future. There are tests going on with IDPR (Inter Domain Policy Routing). One of the IPv7 proposals (PIP) even has Policy information inside a packet. In the BGP session there was some discussion on "BGP communities" which would let you give a route a certain community (for instance "educational") The CIDR proposal and policy based routing may conflict in terms of optimal address assignment. It was therefore suggested that different policies should have different CIDR blocks. references: internet-drafts: draft-ietf-idpr-* Whois ===== WHOIS is now an official Internet Standard. There will be some effort to recommend a certain standardized output from whois servers. How this will proceed is not very clear to me. Operational Requirements Area Directorate ========================================= The ORAD will restructure itself. They have made a list of actions which will all be reviewed. The intention is to set up a ORAD board of about 12-16 people to review each internet-draft on operational impact. contact person: Bernhard Stockman <boss@ebone.net> Phill Gross <pgross@ans.net> Audiocast/Videocast =================== As you may have heard, parts of the IETF have been audio and videocast throughout the world. The current estimates are that over 200 people joined the audiocast during the week, and some 100 joined the videocast. At this meeting the videocast was one way only, but work in underway to make the videocast two way at the next IETF. This videocast test used 4 frames/sec video using some 200 Kbits/sec if there was motion. For the videocasting of slides, only when the slide changed the 200 Kbit/sec was needed. The audiocast was already two-way, and of good quality. The RIPE NCC joined the audiocast in the middle of the week, and setting up of all the software took about an hour. The audiocast used from the IETF used 64 Kbit/sec, if there was any sound. The NCC has done some tests, and a different encoding which uses 32 Kbits/sec gave almost the same quality. Other encoding mechanisms use only 4.8 Kbit/sec and are still reasonable. The RIPE NCC has proposed an audiocast test within the RIPE community. This could be used to audiocast parts of RIPE meetings, have working group audio conferences etc. contact person: RIPE NCC <ncc@ripe.net> Other topics ============ The following "hot" topics were also discussed, but I did not attend the sessions or have not much to say about them. - Mobile IP, presentations and discussions, approved charter. Steve Deering <deering@xerox.xom> - OPSTAT, finalized draft, ready for informational RFC Bernhard Stockman <boss@ebone.net> - User Connectivity Problems, discussion about ticketing systems Dan Long <long@bbn.com> - DNS, DNS MIB under discussion, restructuring of .com and .edu domains under discussion. BIND 4.9 will come out within 1 to 2 months. It will contain all of the patches people currently use in 4.8.3 plus the RP resource record. Also processing speed and various other bugs have been improved and fixed. Next Meetings ============= 25th IETF - Nov 92 Washington DC (hosted by Sprint) 26th IETF - Spring 93 Columbus Ohio 27th IETF - Jul 93 Amsterdam (hosted by SURFnet) <<=== (A "hands up" poll showed that around 70 percent of all people present at the IESG plenary -- around 250 people -- will attend the first IETF in Europe. All in all, Erik Huizer from SURFnet estimates to have around 500+ people at the Amsterdam IETF)