Re: [mif] Route option for DHCPv6 - next steps?

[Behcet Sarikaya <[e-mail]>]

Hi Lorenzo,

On Wed, Apr 25, 2012 at 10:41 PM, Lorenzo Colitti <[e-mail]> wrote:
> On Thu, Apr 26, 2012 at 05:43, Alexandru Petrescu
> <[e-mail]> wrote:
>>>
>>> In one case, you need to modify it to say "must implement a DHCPv6
>>> route option". In the other, you need to modify it to say "must implement
>>> RFC 4191". Note that the RFC already says that "nodes that will be
>>> deployed in SOHO environments SHOULD implement RFC 4191", so RFC 4191 is
>>> likely already implemented.
>>
>>
>> What does that "nodes" mean?  In that RFC 6204 context I guess it means
>> all entities in the SOHO except the CPE.
>
>
> "Node" is defined by RFC 2460 as "a device that implements IPv6"
>
>>
>> At most, I think it means that the CPE sends 4191 RAs to SOHO Hosts which
>> neead to read 4191 RAs.  I don't think it means a CPE router to read 4191
>> RAs sent by CPE+1 ISP routers.
>
>
> Nope. The CPE is a node, and thus per RFC 6434 (IPv6 node requirements) it
> SHOULD implement RFC 4191 if it's deployed in a SOHO environment.
>
>>
>> A router to read 4191-specific-route does not exist today.
>
>
> The linux kernel supports it, I believe.
>
>>
>> In this case, what would one prefer to specify - a 4191 router to read
>> specific routes from 4191?  Or a DHCP Client already doing Prefix Delegation
>> (a Requesting Router) to read DHCP route options and default route options?
>
>
> I think RFC 4191 has much richer semantics (multiple sources of information,
> early deprecation, deprecation when the router originally crashes). So I
> would prefer RFC 4191. As we know from this thread, others disagree. :-)


RFC 4191 needs to be extended in view of route option work in MIF, I
think. This is what we have done in
http://tools.ietf.org/html/draft-sarikaya-mif-6man-ra-route-00

Take a look.

Behcet