Re: new draft on IPv6 CPE router available for review

[Shin Miyakawa <miyakawa@nttv6.jp>]

Mikael,

I see. Then, next time, please refrain from using a bit offensive words such as

> If you want to remove the merit of my proposal on the basis that I haven't 
> supplied RFC quality text to support it, well, then the IETF process is 
> broken because it doesn't listen to what the operational community has to 
> say. 

This sounds that YOU ARE the operational community ITSELF.
But now I understand, it is not what you meant, but please understand it sounds

.......

> Exactly, which is why I have a problem with it. I want the option of not 
> assigning a WAN IPv6 address. 

OK. You want the OPTION now. Sounds very nice.

> > I see. OK Then, how about like this ? (This is actually just an idea. I think
> > that Wes and Hemant are able to write better text than mine...)
> >
> >    The CPE can acquire global IPv6 address for WAN interface using SLAAC or DHCPv6.
> >    In this case, The CPE can be anything, a router CPE, a host, strong host model or
> >    week host model, following RFC3484 recommendation or even not.
> >    If operator that the CPE is going to connect, does not assign
> >    any global IPv6 address for WAN interface, there is a limitation on CPE
> >    architecture so that source address on it should be selected from
> >    PDed address which is not recommended in RFC3484.
> 
> This writing is technically correct in aspect to what I want

Very Good. I really appreciate that we are coming closer to achieve a consensus.

> but the writing (at least to me) implies that not following RFC3484 in the "global 
> IPv6 WAN address" aspect is a bad idea, which I think is a bit biased.

fmmm.. I see..

Of course, it still seems to me that RFC3484 recommends to select address
from the addresses which are assigned to the WAN I/F.
and some existing implementation (such as MS Vista) do this so...

> If you cut out "which is not recommended in RFC3484." then it becomes more 
> neutral. 

But, if you think this is a matter of controversial, it is OK for me to cut out
"which is not recommended in RFC3484"  from above text. Let us do so.

> I don't really understand why you so strongly oppose my idea? 
Of course, I did not really understand why you so strongly opposed my idea too :-p
             ===                                                ==

OK.

If your idea is allowing the link-local only model is an option, then fine for me.
                                                      =========
If your idea is the link-local only model is the only choice, then I oppose it.

Please understand that I "prefered" (and actually still prefer) your idea originally, 
Otherwise we did not write RFC4241 that contains following text

----------- from RFC4241 --------------
>2.4. Address Assignment
>
<snip>
>   Because a link-local address is already assigned to the CPE's
>   upstream interface, global-scope address assignment for that
>   interface is optional.
----------- from RFC4241 --------------


But now, because I think that we should allow all the implementation
which is compatible with today's standard set can be compatible with the service
(This's the STANDARDIZATION, isn't it ?),
the SERVICE should be ready to assign global IPv6 address for WAN interface of the CPE.


> If the ISP highly values to have IP separation of customer IP space and 
> ISP IP space, then they can deploy along my proposal,

maybe. or maybe not.

But if the ISP highly values to allow any implementation which is compatible 
with IPv6 standard set to be used by its customers, 
then they will not deploy along your proposal.

Best wishes,

Shin Miyakawa