Re: [DNA-BOF] Detecting Network Attachment BoF description (rev 2)

[Alper Yegin <alper@docomolabs-usa.com>]

>>>> Or, host might change its access point and consequently access router, but
>>>> the two routers might be advertising the same prefix. In that case, IP
>>>> address of the host does not change, but default router information must
>>>> change. So, even if the subnet does not change, host has to react to
>>>> this.
>>> 
>>> Does it? I assume that the AR is the default router on the link. If
>>> the AP changes, but the AR does not, why is this visible to IP? It
>>> doesn't need to be.
> 
>> Imagine both the access point and the access router changes. But the
>> previous and new access routers are in the same multi-link, hence they are
>> advertising the same prefix. Host has to change its forwarding table and
>> neighbor cache by adding the new default router. But host does not have to
>> change its IP address.
> 
> In what way and why?

Because the link prefix is the same, the IP address configured on the
previous link can be preserved on the new link.

> 
> Note: I don't know what a mulitlink subnet is, as we don't have a
> crisp definition, so I don't know what its properties or the
> implications of a node moving from one part of the multi-link subnet
> to another one.

All I know about this is in (now expired)
draft-ietf-ipv6-multilink-subnets-01.txt .

I could find the first version.
http://www.watersprings.org/pub/id/draft-ietf-ipv6-multilink-subnets-00.txt
I'm CC'ing the authors....


Jumping to your other message below:

>> > Is this what you meant?
>> > If so, should we be thinking about multi-link subnets?

>> Yes, why not? Unless IETF decided to not adopt
>> draft-ietf-ipv6-multilink-subnets-01.txt... This is an expired draft, that
>> was once an official IPv6 WG item. Later it was decided to tackle this
>> elsewhere, and I don't know its latest status...

> At present, it is unclear that "multilink subnet" work is going
> forward in the IETF. It is NOT being worked on my any current WG,
> though it is being discussed a bit again within the zerouter context
> (which is not a WG yet). Thus, I'm not sure how productive it is to be
> desinging around a moving target of unknown properties...

If this is the current state, I agree. I was under the impression that IETF
was doing this work, because of this wording in IPv6 charter:

    o Develop Proxy Router Advertisement solution for prefix
    delegation and publish. This enables a simple site border router to
    re-advertise downstream a prefix it hears on its upstream link.
    The Multi-Link subnet work will be used as the basis for this.
    Note: General multi-link subnet work will be done elsewhere
    in the IETF. 

>> Not really. Because even if the subnet is the same (if we are defining the
>> subnet based on the prefixes) there might need to be changes (see earlier
>> discussion on multi-links).

> Can you give an example  of some of those changes, and why they would
> need to be taken into account?

If the physical link changes, but not the prefix used on the link, the host
still has to change its default router entry because now it is connected to
another router (that is likely to have a different link-layer and IP
addresses). This has to be taken into account by the host if it is in a
multi-link subnet. Whether this should be taken into account in this WG is
another story.

Alper