In message <[email protected]>
RJ Atkinson writes:
>
> On 30 Sep 2012, at 19:33 , Curtis Villamizar wrote:
> Claim (A):
> > Breaking SLAAC is bad.
>
> Breaking SLAAC breaks interoperability, and will keep
> various networked devices disconnected, which is a
> *fundamental* problem.
>
> Fundamentally, DHCPv6 is not widely enough supported
> (today / soon) for it to be a sufficient answer
> to getting network-capable devices onto the network.
>
> There might be a time in future where the situation
> with DHCPv6 support changes, but we aren't there today,
> and we won't be there soon.
>
> Claim (B):
> > Consumer oriented providers handing out /64s
> > to home nets is also bad.
>
> Agreed (s/consumer-oriented/any/).
>
> HomeNet WG ought to be VERY clear about this.
>
> HomeNet WG ALSO ought NOT "enable" or "encourage"
> such behaviour by encouraging ANY scenarios
> where SLAAC can't work properly.
>
> Claim (C):
> > But if a homenet does get nothing more than a /64,
> > and multiple subnets are needed, then SLAAC gets broken.
>
>
> I don't think that last bit above is broadly agreed,
> at least as of now.
>
> It does not obviously follow that "keeping devices
> disconnected" is better than "not being able to subnet
> as much as desired".
>
> Separately, it is not obvious that those are the *only*
> two possible approaches to the issue you seem concerned
> with (i.e. /64 from a provider && desire subnets).
>
>
> SUMMARY:
>
> >From my perspective, there is a logic fault in the
> combination of the quoted statements. The quoted
> text is equivalent to logic of this form:
>
> If (A && B) THEN C.
>
> However, from where I sit, the combination of (A && B)
> does not necessarily imply C. It could mean that one
> decides to waive the requirement (B) to keep every device
> connected, or that one devices a new alternative (D),
> or something else entirely.
>
> If there are some additional data that would make the
> above true, then it would help to get those onto the
> table, and figure out how they would factor into the logic.
>
> Cheers,
>
> Ran
Ran,
You are suggesting if A then !B. I prefer if A and B, then C.
Your proposed if A then !B implies:
1. A guest subnet cannot be supported.
2. A separate legacy device subnet with IPv4 NAT can't be
supported.
3. A subnet for delay sensitive local audio and/or video devices
can't be supported.
4. Delay and jitter sensitive devices such as SIP phone can't be
put on a separate subnet.
5. All WiFi and LOWPAN and GbE segments must be bridged together
ragardless of preformance penalties.
6. Performance pigs can't be isolated.
I think these were all objectives we would like to support. If
supporting this means splitting up a /64 if A, then I'm for it.
If some device can't live without SLAAC, then the consumer has two
choices. Either 1) bridge everything together, or 2) complain until
you get a firmware upgrade or get rid of that device.
I'd pick 2 above, but I'm not the average consumer and neither is
anyone on this list. I suspect many of us (but not most) already have
subnets in our homes, and do so for good reasons.
Curtis
_______________________________________________
homenet mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/homenet
