RE: Open issues list? [Re: New (-02) version of IPv6 CPE Router draft is available for review]

["Hemant Singh (shemant)" <shemant@cisco.com>]

Pekka is correct.  It has been discussed in this mailer during
discussion of this draft that link-local's will never work in all cases
for every app.

A CPE Router, say rtr1, can detect another CPE Router, rtr2, cascaded
behind it, if the DHCPv6 server on rt1 receives a DHCPv6 message for
IA_PD.  Since the domain is a home, a rouge router in the home can spoof
such behavior, in which case the detection may be unreliable.

Hemant

-----Original Message-----
From: Pekka Savola [mailto:pekkas@netcore.fi] 
Sent: Wednesday, July 30, 2008 3:53 PM
To: Iljitsch van Beijnum
Cc: Hemant Singh (shemant); Brian E Carpenter; v6ops@ops.ietf.org; Alain
Durand
Subject: Re: Open issues list? [Re: New (-02) version of IPv6 CPE Router
draft is available for review]

On Wed, 30 Jul 2008, Iljitsch van Beijnum wrote:
> ULAs are only really needed with nested CPEs, because if there's only 
> a single CPE you can use link locals.

The first part of the sentence is debatable.  The last part is wrong. 
Many applications just don't work if you use link-local addresses. 
(As a side point, could one CPE even reliably know if there are other
CPEs?)

FWIW, my belief is that the link-local addresses should never need to be
used by humans who can't describe the TCP connection establishment
procedure.  The implication of what you suggest would imply otherwise. 
The pain with multi-interface hosts and making apps deal with scope
indexes is just too great.  I personally don't care for ULA that much
myself, but if the alternative is to try to use v6 w/ link-local
addresses, ULA at least on surface seems like a better approach.

-- 
Pekka Savola                 "You each name yourselves king, yet the
Netcore Oy                    kingdom bleeds."
Systems. Networks. Security. -- George R.R. Martin: A Clash of Kings