[Lista ArNOG] Sobre el mal diseño de SLAAC y DHCPv6 (Fwd: Re: [v6ops] New I-D: SLAAC and DHCPv6 (Fwd: New Version Notification for draft-gont-v6ops-host-configuration-00.txt))

[Fernando Gont]

FYI: <https://goo.gl/tc8qJw>

Resumen del problema a manos de Brian Carpenter (de esos poquitos que
vale la pena leer).

P.S.: Y pensar que te venideron que unos de los aspectos maravillosos de
IPv6 es que es "plug&play".... que la inocencia te valga! ;-)

Saludos,
Fernando




-------- Forwarded Message --------
Subject: Re: [v6ops] New I-D: SLAAC and DHCPv6 (Fwd: New Version
Notification for draft-gont-v6ops-host-configuration-00.txt)
Date: Wed, 1 Mar 2017 11:30:27 +1300
From: Brian E Carpenter <brian.e.carpenter en gmail.com>
Organization: University of Auckland
To: Fernando Gont <fgont en si6networks.com>, james woodyatt
<jhw en google.com>, IPv6 Operations <v6ops en ietf.org>

On 01/03/2017 10:12, Fernando Gont wrote:
...
> If your network has such a requirement, use RDNSS rather than SLAAC.
> 
> We're not mandating what you must *use*. Actually, we want to allow you
> to make that choice. Right now, you can't. Because Google's Android
> refuses to implement stateless DHCPv6, and MS Windows and a number of
> router vendors refuse to implement RDNSS.

This is the point. We are supposed to care about interoperbility, and
this is a breakage in interoperability specification, which has led to
breakage in real-world interoperability. Our bad.

There's another closely related breakage, which is the absence of
default router signalling in DHCPv6. That means that even if a site
must use DHCPv6 for configuring stuff that RAs cannot configure, they
must also use RAs for configuring stuff that DHCPv6 cannot configure.

Something like bad systems design.

    Brian