Re: I-D Action:draft-ietf-v6ops-ipv6-cpe-router-03.txt

[Eduard Metz <etmetz@gmail.com>]

On Fri, Jan 8, 2010 at 4:10 PM, Wes Beebee (wbeebee) <wbeebee@cisco.com> wrote:
>>>   WPD-1b:
>>>
>>>           (a)  The IPv6 CE router MUST support DHCP prefix delegation
>>>                requesting router behavior as specified in [RFC3633]
>>>                (IA_PD option).
>>>
>>>           (b)  The IPv6 CE router MAY indicate as a hint to the
>>>                delegating router the size of the prefix it requires.
>>>                If so, it MUST ask for a prefix large enough to assign
>>>                one /64 for each of its interfaces rounded up to the
>>>                nearest nibble and MUST be configurable to ask for more.
>>>                The IPv6 CE router MUST be prepared to accept a
>>>                delegated prefix size different from what is given in
>>>                the hint.
>>
>>This may be getting a bit to fuzzy. I'd prefer something more along the lines of the original statement.
>>Assuming the intention is to ensure that each IPv6 interface in the CE router can be configured with a
>> /64, should the requirement not be that the CE router does not request (or hint for) a prefix that is
>> too small? (without going into saying how large it actually should be)
>>
>> Maybe this has been discussed before, but what is the desired behaviour of the CE router in case the
>> delegated prefix actually is too small to assign a /64 to each of its interfaces?
>
> I suspect that the behavior for a too small prefix would be vendor-dependent as I don't think that we could end up with recommending a specific behavior that would pass IETF consensus.  However, I think that asking for a prefix large enough to do SLAAC, even if you don't get it, is behavior that is completely reasonable and the right thing to do - but shouldn't be mandated.  Stating this hint in this document is a heads-up to service providers to expect that some CE routers may have this behavior.  Therefore, I agree with the text of WPD-1b as is - subject to possibly breaking up MAYs and MUSTs as we're doing elsewhere in the document.
>

So it's a hint that some devices may provide hints ;-) .. this may be too subtle

I'd suggest to make it a more concise requirement. The first two parts
to me are the core:
The first part of WPD-1b describes the option of providing a hint on
prefix size, this is already in RFC3633, but it can be restated here.
The second part of WPD-1b relates the prefix hint to the CE
interfaces. With regard to this I meant to say that -if- the CE router
provides a hint, it should not hint for a prefix smaller than required
to assign a /64 to each of its interfaces. So I think we agree here.


The rest I think can (or even should) be left out:
The third part seems to suggest a default value for hinted prefix
size, based on number of interfaces, or manual configuration. This
could be included as an example of how the prefix size could be
determined, I do not think it should be part of the requirement.
The last part of -1b is already described in RFC3633, for clarity may
be good to restate, but not necessary.


A question: is WPD-1b meant to apply to all DHCP messages in which the
prefix length is specified by the requesting (CE) router? Or just the
case of initial boot (no previously assigned prefix). If a prefix
previously was assigned, I assume the CE router should include this as
a hint, and not provide a hint according to WPD-1b?

/Eduard

> - Wes
>