Intarea Minutes IETF 97 Seoul 16/11/16 15:20-16:20 Chairs: Juan Carlos Zuniga (Sigfox) (JCZ) Wassim Haddad (Ericsson) (WH) Minutes - Ian Farrer (IF) 1. Agenda Bashing, WG & Document Status, Announcements (Chairs) Mark Townsley (MT) - draft-ietf-intarea-tunnels Update (No slides) MT: There will be another rev and then we think it's ready for WGLC JCZ: OK Lucy Yong (LY): One section regarding implementation summary is empty. Why do you think it's ready for WGLC? MT: Maybe we'll end up removing it. I don't know. If there's outstanding issues please let us know. Erik Nordmark: Section 5.5 - there are headings but no content. MT: We'll end up removing them. LY: The more concrete part is MTU a MT: The ones that are empty will probably be removed. Wait until we get to the next rev. It's been a long time and it needs to move forward. JCZ: Please send your comments to the list ---- 2. Announcement - IEEE 802.1CF OmniRAN, Juan Carlos Zuniga ---- 3. GUE and Extensions, Tom Hebert (TH) draft-ietf-intarea-gue-00 draft-herbert-gue-extensions-01 draft-herbert-nvo3-ila-03 MT: Maybe the BIER stuff would work for multicast. It does replication in a different way TH: It still uses multicast addresses? MT: No TH: Then it might work MT: Do you see active migration of containers as something that FB wants to do? TH: Yes. We have a lot of migration, if we have to kill task, we loose progress. There's other reasons. Most of this is implemented, but for a full scale Data Center it's huge. The earlier advantage of ILA is an address per task simplifies lookup. Every task has an address. We never get port conflicts. Rolf Winter (RW): In NVO3 they decided to halt until a single solution is published. Does this apply to GUE? Are you moving it to Intarea? TH: It's a generic protocol. GRE hit a wall. RW: The question is will the work be halted due to NVO3. TH: I don't think they are pursuing an existing solution. They are looking for a new one. Suresh Krishnan (SK): The answer is no. I went through the draft. There's no scoping. The way it's written today, it can't work on the Internet. You need to scope it to move forward. TH: I'm not sure that is true. We're looking at it as a solution for 5G mobility. Think of them as tunnels. When packets enter the carrier, we can do translation and get them to the right base station. If 2 carriers are talking to each other,there's enough information to get it all the way to the destination. ILA is only needed at the endpoints. JCZ: Final question. Are you assuming that the draft will worked on here? Bob Briscoe (BB): I reviewed it's a long doc. It'll take a while. ---- 4. IP Broadcast Considerations, Rolf Winter (RW) draft-ietf-intarea-broadcast-consider-01 Tim Chown (TC): Does your document cite the work that Christian and Daniel are doing in DNSSD? RW: No TC: There's 2 docs, 1 deals with privacy issues, the other with pairing using obfuscated names and TLS. It would be useful. I'd encourage you to reference it. ---- 5. Extended Ping, Ron Bonica (RB) draft-bonica-intarea-eping-02 RW: A regular ping is nice because you reach the address you pinged. RB: Do you have any guarantee that the ping actually traversed the pinged interface? RW: The info is the same that you would get from the CLI. RB: The only thing that's different is regular ping tells you the protocol is up. This tells you about other protocols on the interface. RW: You can try and figure out all of the interfaces that are up on the box. RB: You can say that I will only honor pings by mac address or v6address etc. RW: It's a security consideration. RB: It's in the security consideration section of the doc. RW: Wouldn't it be better not to send an error, rather than saying why you didn't get a reply? This could be a security concern. RB: Let's discuss it more off list Carlos Bernados (CB): Do you want to filter on what you provide a reply. RB: As the draft is written now, you can get all of them. Bob Hinden (BH): This works on one node, your sending it to a loopback and getting information on an interface. It would be useful if you could use one node to probe another node that doesn't have global scope. RB: I thought about it, it would make things much more complicated. BH: Not much would have to change. RB: Let's discuss it offline. It's interesting. BH: You'll need to pass this by 6man. RB: Absolutely. Jen Linkova (JL): I'm trying to understand the use case. I don't, but I can see that others might. If you're specifying LL you need to specify zone ID. LL without zone ID is useless. RB: Good point. JL: You shouldn't just call it source and destination. RB: I will put that in the doc. JL: I don't like the way that you say it's just like ICMP, you don't return the whole message. RB: That's true, yes. RB: Is there enough interest for CFA? JCZ: We'll take that to the list. RB: I'll update the draft with the comments. ---- 6. IP over intentionally partially partitioned links, Erik Nordmark (EN) draft-nordmark-intarea-ippl-05 No Comments JCZ: We can make a call for adoption on the list. ---- 7. Special Purpose IP Address Registries, Ron Bonica draft-bhcv-rfc4890bis JL: I tried to read it, but couldn't find it. JCZ: It's AD sponsoered. SK: There's a typo in the slides, it's 4890-bis, not 6890-bis. I can AD sponsor it. Ralf sponsored the last one. I'm just going to run it through this group. BH: This is the same document that was discussed in Berlin ---- 8. Multiple Access Management, Satish Kanugovi (SK) draft-kanugovi-intarea-mams-protocol-01 David Black (DB): How does this relate to the work in the Banana BOF? SK: It relates pretty much. DB: That's where this draft belongs. SK: We are presenting this tomorrow. ---- 9. Interconnecting Network Sites by IP Tunnels, Lucy Yong (LY) draft-yong-intarea-inter-sites-over-tunnels-00 (out of time) ----