Re: DHCP failures (was RE: Do you want to have more meetings outsi de US ?)

[JINMEI Tatuya / 神明達哉 <jinmei@isl.rdc.toshiba.co.jp>]

At Fri, 3 Aug 2007 07:16:50 -0400,
Ralph Droms <rdroms@cisco.com> wrote:

> The dhc WG has received a request to develop, on behalf of network  
> operators who have expressed the specific requirement that they do  
> not want the operation of their network to depend on RAs, new options  
> for DHCPv6 to pass prefix and default router information to a host.   
> That information would allow a host to get all of its configuration  
> information from DHCPv6, obviating the use of RAs.
> 
> Will the IETF elite allow the standardization of these new options to  
> support this different mode of configuration or will those network  
> operators who would prefer to use only DHCPv6 be forced to continue  
> to use RAs?

(I'm not trying to make a specific opinion about the RA-only vs
DHCPv6-only argument)

It would help if you could provide more details about the requirement:
who are exactly the operators wanting DHCPv6-only operation, why they
want to do it, what's their (technical) problem in using RA to get
configuration information that is currently provided only by RA, etc.

I doubt one can convince others to introduce something new that can
already be done by an existing feature simply because there is someone
who wants it; the important point is that why they need it and whether
it's technically reasonable.

BTW, I see one reason that we might need to provide on-link prefixes
via DHCPv6 even though RA already has this functionality: if we
operate a closed, single-link network which contains a DHCPv6 server,
a local DNS server, and any other servers that provide services in the
network, but no router, then the nodes have to communicate using
either link-local addresses or global addresses assigned via DHCPv6.
And if the nodes want to avoid link-local addresses due to its
scope-related difficulties, they need to detect other global addresses
are on-link.  However, since DHCPv6 does not provide that information
and the "on-link by default" rule is going to be deprecated in
2461bis, there is going to be no standard way for such nodes to get
the prefix information.  Note that this problem didn't exist when the
DHCPv6 protocol was designed thanks to the "on-link by default" rule
as a last resort; but now that the assumption does not hold, I think
the DHCPv6 design principle can also be revisited.

On the other hand, I personally don't see the case where we need to
provide the default router address(es) via DHCPv6.

					JINMEI, Tatuya
					Communication Platform Lab.
					Corporate R&D Center, Toshiba Corp.
					jinmei@isl.rdc.toshiba.co.jp