Minutes IETF106: v6ops
minutes-106-v6ops-00

Chair slides
Fred Baker mentioned note well, and that expected presentation from RENU had to
be canceled because presenter had to defend thesis.

=======================
Neighbour Cache Entries on First-Hop Routers: Operational Considerations
Jen Linkova presented slides:
https://datatracker.ietf.org/meeting/106/materials/slides-106-v6ops-neighbour-cache-entries-on-first-hop-routers-operational-considerations

Dave Thaler: Both have disadvantage. Can you consider a 3rd option where you do
neighbor discovery on receipt of the first packet? Jen: Signaling control plane
from data plane is maybe not the best idea. Jen: If we can do everything in
6man... Fred: How would you approach that. Warren Kumari: Bob Hinden (as 6man
co-chair): If the v6ops WG accepts this then one should be preferred. Fred (to
Bab): Is Jen on your agenda? Bob: Yes. Jen: Do we want to have a discussion
about all options and pick one to take to 6man or take all to 6man? Fred: I
want to leave all discussion of protocol to 6man. Warren: Document here is to
determine if there is need for a solution. Jen: We agreed last time that there
is a problem. Warren: I think it would be good to ask the room if they think
this needs solving. Fred: We did hum last time and it was loud. Fred: We can do
another hum. If you think this needs solving, please hum? <loud humming> If you
disagree? <silence>

========================
Default IPv6 Local Only Addressing for Non-Internet Devices
Mark Smith presented slides:
https://datatracker.ietf.org/meeting/106/materials/slides-106-v6ops-default-ipv6-local-only-addressing-for-non-internet-devices

Jen Linkova: If the device doesn't get any GUA address, now it needs a ULA
only. What about networks with multiple IPv6 networks? Now they need ULA and
GUA for this device. Mark: Smart Meter manufacturers build devices NOT to be on
the Internet. Jen - Mark - Jen - Mark

Fernando Gont: Is there any requirement of the CPE router?
Mark: I realize that I'm effectively requiring
Fernando: If device were to configure GUA and ULA, you can have a device with
both, and applications that need ULA could just bind to the ULA. Mark:
Fernando: What has to be on the interface? Mark: You might have a light bulb
devce that says even though it can do certain things, it won't, because it's
just a light bulb. Fernando: The application should ... Dave Thaler: The thing
I worry about is when a printer only has a ULA and the device looking for it
only has GUA. Mark: I do think about that. Éric Vyncke: Mark: We're trying to
be incrementally better. Jordi: I'm not sure it's so obvious to the vendor
whether it needs to get on the Internet or not, it may depend on where you
install that device at a given time. I think this is operationally complex. If
we want this as a security tool, let's do that at the device firewall level or
CE firewall or both. Andrew : I think we could do better security on CPE and
edge. I don't trust ... Mark: I don't trust the firewall in my router. David
Somers-Harris: Putting on a non-technical user hat... People often don't
understand features like "print from the cloud" and understand why it doesn't
work because it only has a ULA.

===============================
464XLAT Optimization for CDNs and Caches
Jordi Palet presented slides:
https://datatracker.ietf.org/meeting/106/materials/slides-106-v6ops-464xlat-optimization-for-cdns-and-caches

Fred: Comments?
Andrew: Considering everything that was said in Montreal about A records, what
has changed since then? Jordi: I think that was for the other part. And I think
we have resolved that issue. My suggestion is that we need to go to another
draft for the fake A records. Jared Mauch: DoH is shifting DNS away from ISPs.
How does that impact this? Jordi: You will not get advantage of this, but it
won't break things. Chongfeng Xie: Jordi: Those services are not yet
IPv6-enabled. Chongfeng Xie: Jordi: But that's my point. The closed service is
not IPv6-enabled so it has no impact. Fred: An enterprise will often have their
own DNS and policies and are concerned about DoH and DoT. I think you need to
check again if the enterprises are concerned. Jordi: If you are not using  the
optimization, you are not breaking anything.

================================
IPv6 Point-to-Point links
Jordi Palet presented slides:
https://datatracker.ietf.org/meeting/106/materials/slides-106-v6ops-ipv6-point-to-point-links

Fred: Comments? <there were none> Thank you.

===============================
Reaction of Stateless Address Autoconfiguration (SLAAC) to Flash-Renumbering
Events Fernando Gont presented slides:
https://datatracker.ietf.org/meeting/106/materials/slides-106-v6ops-reaction-of-stateless-address-autoconfiguration-slaac-to-flash-renumbering-events

Barbara Stark: I support moving forward.
Jen Linkova: I think this is good.

==================================
Improving the Reaction of Customer Edge Routers to Renumbering Events
Fernando Gont presented slides:
https://datatracker.ietf.org/meeting/106/materials/slides-106-v6ops-improving-the-reaction-of-customer-edge-routers-to-renumbering-events

Jordi: I think you are too optimistic that CPE router vendors will take these
recommendations. I think this would increase the price of the devices, so they
won't. I think we should go for a different solution, including telling the
ISPs to use IPv6 in a different way than IPv4. Fernando: But maybe some CPE
router vendors will want to do this. Andrew: I like what I see here. I don't
think the fact that some won't do this is a reason not to do this. This gives
me something to point at when I want vendors to do this. Tim Winters: For how
long would you do this? Valid lifetime? Fernando: Yes. Jen: I think it does
make sense to certify. Jordi: I'm not saying we shouldn't do this. But I think
we should also try to fix from the other side. We also need better tools from
the ISP side. Fred: Put something in a RA to indicate "whatever I told you
before is now invalid". Fred: Thank you, Fernando.

Fred: In past years we found it helpful to poll people what they thought of a
draft. Andrew: Bernie Volz: You wouldn't want to use a single bit because it
can't provide enough info. Mark Smith: I'm skeptical of the effectiveness of
trying to solve this problem on the CPE router. Fernando: It's not that we
think a solution only belongs in one place. We're just trying to provide
solutions wherever it makes sense.

Dmytro Shytyi : I think your proposal could be applied to other cases.
Fernando: I agree the document could be more general and not just about CPE
router.

Fred: Between now and IETF 107 we should try to make progress. I want to use
first 3 weeks in December and first 3 weeks in January to ask questions leading
to progress on these drafts. Please feel free to say something when I send
those emails.

Dmytro Shytyi : Is it possible to present something?
Fred: There is time if you can give it to me.
<did not succeed in time>