Re: Adoption Call for <draft-troan-6man-universal-ra-option>

[Philip Homburg <[e-mail]>]

>The RA / DHCPv6 split we have today (excepting addresses in DHCPv6, and DNS
>in RAs) aligns with both the system and network role boundaries and the
>Internet protocol layer model.

There seems to a notion promoted by the proponents of RAs that hosts
do not need unique, per host configurable default routers. Obviously,
in it's current form, RA cannot provide that. 

At the same time, there are networks that a decade or longer have had the
same IP address as default router. Somehow proponents of RA claim that it
is very important to have fate sharing. Even though in practice this does
not offer anything to hosts. The fate sharing provided by a router not sending
RAs happens on a time scale (tens of minutes) that is not suitable for
today's environment. In practice, hosts have to rely on reachability 
etection in ND to determine if a default router can be used or not.

So instead of creating operational flexiblity where operators can decide how
to run their networks, the result of this working group seems to be to
try to kill all concepts that don't fit into the RA mold.

In don't think this draft is a solution though. We need well thought out and
well written standards. Not an openended system where the real protocol is 
just a custom created by dominant parties.