Please see below for the first draft version of the minutes of the IPv6 wg session during RIPE40. Please take a look at it and report typos, errors and omissions. The minutes (with possible minor corrections) will be declared final if I receive no comments and/or only minor typographical errors by Fri Dec 21. Thanks, David K. --- Minutes IPv6 working group RIPE 40, Prague, October 2nd, 2001 Chair: David Kessens Number of participants: 83 participants A) Status of 6Bone (David Kessens, presentation available from http://www.kessens.com/presentations/) - Noted growth in IPv6 sites - Better understanding of the system - Growing number of inet numbers - More addresses are being allocated - Addition of two countries - 'co' & 'cu' (presumably Colombia and Cuba) - Number of queries are up - No questions were asked B) Addressing on cable systems (Pascal Julienne) - Reminder: example is a typical site in one city/TV and phone connected to one server - Addressing scheme of cable networks are more complicated when connecting cities i. Routable sites are more complex ii. One WAN for one city, one WAN for another, and so on. iii. As you grow cable networks in a country, you can tie up several cities with different servers, making routing scheme more complex - Why is IPv6 attractive for cable networks? i. Unique address requests from large number of end users ii. IPv6 Anycast address may provide effective web traffic between service points - Host address i. IPv4 vs Mac based addresses 1. Idea is not to renumber but use existing addresses ii. Digital TV - Shows service-based aggregation scheme/aggregate by service i. Two possible ways a. Prefer services over geographical areas b. Geographical areas over service In one case, you can manage all services from one department In the other case, you can manage different cities c. Merged scheme has yet to be examined thoroughly - IPv6 vs IPv4 interaction i. IPv6 or IPv4 or use of a tunnel? - Protocol stack i. Multicast aspect in cable network ii. Most countries use DVB today iii. Modifications need to be made for the future iv. SMPTE 305M is IPv6 compatible but not best for Europe Q: Is there any IPv6 vendor support for cable equipment ?!? A: - End-user's PCs are ahead - Set-top boxes and head-end need IPv6 implementation - Digital TV and telephony support unknown - Where do we go from here? Internet on TV is not going well in Germany but speaker believes it is possible Q: We have a problem with Anycast. A: Anycast works only in the routers and not in the service Anycast needs to be investigated to overcome problems Q: You can connect to the Internet via the private topology/connectivity of IPv6, cable systems, /48 systems? A: it depends how it is managed What I show is not an answer of how to do it but an example Q: Along the lines of Francis ... you can't give an end-user one IP address..., one customer in Paris... A: this is just an example - I am working with Euro and US cable set box manufacturers. In one month I can give you names of several companies who are developing equipment A: can we get the names now? (laughter) C) Global IPv6 Routing Table (Gert Doering, presentation available from http://www.ripe.net/ripe/meetings/archive/ripe-40/) - This will not be scientific but impressions; something to show and discuss - I will give an overview of the topic - What is the routing table? - Protecting addresses i. One big difference to new address: everyone gives transit to others ii. Everybody gets connectivity iii. Underlying topology is different from IPv4 - Some continents have different structures, leading to some interesting observations i. (Shows Saturdays' numbers) ii. 6BONE iii. Most important for the future is a /35 iv. Shows distribution graph of total number of prefixes since Saturday v. Shows graph separating from 6BONE and registry base vi. Shows new slide of RIRs, AS, and AS paths vii. Overall there are 171 addresses participating in the table? viii.Average number of Hops is so similar via IPv6 vs IPv4 ix. Observations of 'crap' in the table x. Problems of address length and of too many hops xi. Find out what router is causing them and fix xii. Lazy aggregation - Cable & Wireless example xiii.Need to be more careful of what is put into the IPv6 table otherwise it will look like IPv4 table, full of 'crap' xiv. Invalid AS numbers (shows example of Mexican ISP) Comment: I saw AS numbers being stolen (gasps/laughs) - I want to make IPv6 work, so if people use 'crap', I will use that For now, I just observe weird things and call them to make them stop - Conclusions: it works, people are hooking up to IPv6, it is still experimental, people are cooperative, interest is growing, even with the illegalities (stealing, etc.) Comment: on filtering - we can discuss this at next meeting - We need to be careful of which packets are used for certain circumstances - Shows slide of references Comment: We have 250 routes; it would be interesting to have a registry as we did before. - This is something to be discussed later Q: You mentioned what could we do to improve? Anything we can do here (at RIPE?) A: We can look at the table and if we see irregularities; just ask them for the reason; many may not respond; but if everyone just talks to their neighbor.ask if it is done on purpose.. Q: I am tempted to place this on the agenda topic next time. Q: I want to ask people what addressing scheme they use for point-to-point links A: I use 127's. I have never seen any problems using them. D) Discussion on RPSL and IPv6 routing policies (Florent Parent, presentation available from http://www.ripe.net/ripe/meetings/archive/ripe-40/) Florent discussed whether the implications on RPSL regarding RPSL. He discussed the various options of adding attributes and classes. The decision was made to form a maillist to discuss the RPSLng: <rpslng@ripe.net> E) Pan-European IPv6 IX Backbone (Jordi Palet, presentation available from http://www.ripe.net/ripe/meetings/archive/ripe-40/) - Remarks on receiving funding for IX backbone - Began project in April 1983 - Project still under negotiation i. Hopefully, we'll start on December 1st - Correct pronunciation: 'Euro6IX' as in 'EuroSix' - Build a native IP structure i. Want it to be 'open' - Will receive funding from the EU - "Chicken and Egg" problem overcome - Gaining real IPv6 experience in the real world i. Simulates real commercial networks - Euro6IX Goals and Objectives i. Consortium numbers 1. Main Telco's represented by their research arms 2. Universities 3. Sponsors are international ("and it's just starting") ii. Network hierarchy iii.Network map iv. Infrastructure set-up v. Work package summary vi. Research activities vii.Users/User group involvement viii.Access Usage policy 1. For EU funding, we must stay clear of conflicts F) IPv6 in European Region Input from the audience requested: IST/6NET IPv6 testbed (Bernard Tuy, presentation available from http://www.ripe.net/ripe/meetings/archive/ripe-40/) Q: What is the difference between Euro6IX and IST? A: Membership is different Q: But the goals seem to be the same. A: Yes, most are the same. Both projects will incorporate a lot and be common but will never be equal. Money has been put into IPv6 development and that's important.