DHCPv6 and MAC Address inclusion

[Niall O'Reilly]

On 26 Jan 2012, at 22:57, Ted Lemon wrote:

>> They are necessary for DHCP to function in conformance with the standard.
>> > So they need to exist so that they can exist?  Sounds like a circular argument.
> 
> The standard says "do X".  If your implementation does Y, you have not implemented the standard.
> How is this circular?   The DUID identifies the client; without it, you can't differentiate between clients.

	Ted,

	As you've suggested in an earlier message (to <dhcp-users at lists.isc.org>),
	the discussion really belongs on the DHCWG list.  I'm trying to move it there.

	[Anyone replying may find it useful to subscribe to <dhcwg at ietf.org> first.]

	I think you're missing the point.  I agree with the OP that the argument
	is circular.  This is because the context of the standard is self-referential
	and self-motivating, without regard to the need for compatibility and
	interoperability with existing business processes, some of them in place for
	20 years or so.  These quite commonly depend on an identifier which can be
	obtained from a label on the outside of an as-yet-unopened carton containing 
	the device to be registered so that DHCP and other network provisioning tasks
	can be performed.  These tasks typically need to be completed in advance of
	connection of the device, by an otherwise unaided customer, to the net.

	Feature parity between IPv4 and IPv6 matters, and needs to be available 
	not just in the network, but at the business process layer.  Avoiding 
	feature parity, even for an apparently good reason such as discouraging
	non-compliance, is an approach which obstructs adoption of IPv6.


	Best regards,

	Niall O'Reilly
	University College Dublin IT Services