FW: new draft on IPv6 CPE router available for review

["Wes Beebee (wbeebee)" <wbeebee@cisco.com>]

Forwarding for Hemant to continue the discussion...

- Wes

-----Original Message-----
From: Hemant Singh (shemant)
Sent: Tuesday, July 01, 2008 3:20 PM
To: 'Shin Miyakawa'; davidm@thetiger.com
Cc: swmike@swm.pp.se; Wes Beebee (wbeebee); v6ops@ops.ietf.org;
yasuhiro@nttv6.jp; sven.ooghe@alcatel-lucent.com; Hemant Singh (shemant)
Subject: RE: new draft on IPv6 CPE router available for review

Shin,

Thanks for your comments. We agree with (A) below. As for (B), we leave
it to the mailer to decide if it's a MUST for the CPE router to support
the link-local only WAN model or the MUST can be a MAY. As we said to
Mikael's review yesterday, we see no technical problems with the CPE
router supporting the link-local only WAN model. Of course, we totally
want the first model to be a MUST for the CPE router to support - in
this model the WAN interface acquires a global IPv6 address.

Hemant & Wes.

-----Original Message-----
From: Shin Miyakawa [mailto:miyakawa@nttv6.jp]
Sent: Tuesday, July 01, 2008 3:11 PM
To: davidm@thetiger.com
Cc: swmike@swm.pp.se; Wes Beebee (wbeebee); Hemant Singh (shemant);
v6ops@ops.ietf.org; yasuhiro@nttv6.jp; sven.ooghe@alcatel-lucent.com;
miyakawa@nttv6.jp
Subject: Re: new draft on IPv6 CPE router available for review

Hello, David :-)

> I would add my support to Mikael's request that the draft has the 
> option of having a global unicast WAN address OR having link-local 
> only WAN on a routing CPE. Would it be okay to incorporate the content

> of Shin's draft into the CPE discussion/draft?

David, as we have discussed this several months off from this mailing
list in various places on the Earth, you know the reason why I said.

Then.... "OR" means sometime ambiguous.......

Actually, my opinion is

(A) the draft should have "discussions" text to explain why those are
recommended
    and this draft should be more generalized not only for cable TV
systems.

    For example, section 1 in this draft's clearly said that
    "WAN interface is Ethernet encapsulated", but it should be converted
to
    "WAN interface may be Ethernet encapsulated but not limited too" 
    or something like that because we need to think about other
semantics like PPP.

(B) Because a strong host model is one of possible implemenation of CPE,
so

    "Global unicast WAN address" MUST be supported
       and                       ^^^^
    "Link-Local only WAN on a routing CPE" MAY be supported

as the "access concentrator"'s specification which provides the service
to the IPv6 CPE.


Best wishes,

Shin Miyakawa

P.S.

However I also prefer Mikael's model personally, we have to be very
realistic and tolerant to any implemenations compatible with the
standard.