(Thanks to Ian Farrer for taking notes during the meeting!) IntArea WG Agenda IETF 95 Tue, 16:20-17:20, Room Buen Ayre C Chairs Juan Carlos Zuniga (JCZ) Wassim Haddad (WH) Suresh Krishnan (SK) Minutes taken by Ian Farrer Adgend 1. Agenda Bashing, WG & Document Status, Announcements (Chairs) 10 minutes 2. Ad Hoc Wireless Networks, Charlie Perkins 5 minutes draft-ietf-intarea-adhoc-wireless-com-01 3. Multicast over IEEE 802 Wireless, Charlie Perkins 15 minutes draft-perkins-intarea-multicast-00 4. IP Broadcast Considerations, Rolf Winter 10 minutes draft-winfaa-broadcast-consider-01 5. IP over intentionally partially partitioned links, Erik Nordmark 10 minutes draft-nordmark-intarea-ippl-03 6. IP in UDP, Xiaohu Xu 10 minutes draft-xu-intarea-ip-in-udp-03 WH taking over from SK as INTAREA chair SK thanked Brian Haberman for his work as chair 1. Agenda Bashing, WG & Document Status, Announcements (Chairs) draft-ietf-intarea-hostname-practice-00 Christian Huitema (CH) - draft needs to have reviews to proceed. Rolf Winter (RW) - this draft addresses a real problem. JCZ - It's missing reviewers at this point. RW can you do a review? How many people have read it? (1-2 hands) JCZ - We need more people to review. I encourage people to send comments. 2. Ad Hoc Wireless Networks, Charlie Perkins (CP) No questions 3. Multicast over IEEE 802 Wireless, Charlie Perkins Andrew McGregor (AM) - The DMS is still used CP - Can you send text on this? Mikael Abrahammson (MA) - There are two classes of wireless product, consumer and enterprise, this is is not widely in consumer WLAN products. JCZ - MA, your point is valid. It's worth documenting. Donald Eastlake (DE) - The amount of time that it takes to seize the airtime becomes more and more. You can piggyback the multi when you are sending unicast anyway. MA - This is a function on both the AP and the station. The draft should say where a function needs to be implemented as this will affect the adoption time. Erik Nordmark (EN) - Another point having some kind of address registration - the host registers its address. If you do that, it's similar to an ARP cache. There's a draft called: draft-efficient-ip-something Ole Troan (OT) - Do you see that there is work needed in the IETF as well as the IEEE? CP - Yes, they didn't believe there was a problem . OT - There is difference between what's in products and whats in standards. Product often solve this problem in their own way. OT- You've specified the use of an experimental RFC. Does the IEEE specify that ND proxy should be used? I have concerns about this. EN - They may not be doing this. Tim Chown (TC) - when you have a network with lots of nodes, does the performance deteriorate. Is that something you want to consider as well? There's other considerations such as HNCP and DNSSD - In DNSSD the author has said no more than 20 multicast requests a second. There needs to be guidance on this. CP - we're going to have all of the problems and all of the solutions that we know about in the draft TC - it's something that we need to think about. MA - My goal in getting involved was to the the IEEE and IETF to work out a solution between them. There's now a good spirit of interworking. Once these things are figured out, there needs to be guidance in the IETF on how to do this in the future CP - That's a good point and it's with the IEEE/IETF committee. Warren Kumari (WK) - Does anyone think that this isn't a problem? We've improved it here because we've cut down on the broadcasts. Stuart Cheshire (SC) - Thanks, I think this work is important. The number TC mentioned was not pulled out of the air. A full sized frame is about 10k bits, I picked the number for DNSSD based on this. TC - You could mention that if you're running v6 only... (missed) Gabriel Montenegro (GM) - I've heard some some good arguments. We've got a document in 6lo, there are multiple solutions in other wireless groups. CP - All of the solutions that I've heard are really short term JCZ - Please bring these to the list 4. IP Broadcast Considerations, Rolf Winter (RW) CH - Thank you for the data. are typically doing discovery. Ideally we would like devices to use an IETF protocol for this so that they can benefit from our security Did you speak to app developers? RW - It's not just the protocols, it's how they are used. If they still have constant identifiers, then it doesn't help. JCZ - Are you planning to update the draft? RW - There is more data to add. We'll see this next time. 5. IP over intentionally partially partitioned links, Erik Nordmark (EN) No comments 6. IP in UDP, Xiaohu Xu (XX) EN - I don't have any problem with another form of encap. 5405bis - has some wording such as how to handle MTU There's also an encapsulation considerations doc in the routing area workgroup saying that you don't want to use 16 bits, just 14 bits for the ephemeral ports There's something else in the MPLS over UDP draft that is worth considering as well (you might not need this as you're only carrying IP)