Re: [Autoconf] Last Call: draft-ietf-autoconf-adhoc-addr-model (IP Addressing Model in Ad Hoc Networks) to Informational RFC

[Erik Nordmark <[e-mail]>]

On 02/19/10 05:42 AM, The IESG wrote:
> The IESG has received a request from the Ad-Hoc Network Autoconfiguration
> WG (autoconf) to consider the following document:
>
> - 'IP Addressing Model in Ad Hoc Networks '
>     <draft-ietf-autoconf-adhoc-addr-model-02.txt>  as an Informational RFC

I read this draft a few weeks back during the last call. But I didn't 
send the comments because I wasn't up to speed with the WG discussion, 
and I figured I could do that while talking to folks in Anaheim. But 
then the draft was approved.

I have two significant issues with the document.

First of all it seems to conflate the notion of a router ID with the IP 
addresses configured on the interfaces on a router.
Second of all it seems to discourage the use of IPv6 link-locals as the 
IP addresses to configure on interfaces on routers.

But this seems to be counter to the current set of existing well-known 
Internet routing protocols.

For instance, RIPng doesn't even use a notion of router IDs, and is 
required to communicate using IPv6 link-local addresses.

OSPv3 running on IPv6 also is required to use IPv6 link-local addresses 
for the exchanges AFAIK, but the router ID is a 32 bit number.

ISIS has a router ID that is a NSAP address (derived from an IEEE MAC 
address), and doesn't require IP addresses to be configured on the 
interfaces in order to run the protocol between the routers.

Hence router IDs doesn't need to be an IP address, and there is no need 
to stay away from IPv6 link-local addresses for the above protocols. Yet 
this draft has come to the conclusion that things need to be different 
for links with undetermined connectivity, which makes no sense.

Regards,
    Erik