DHCPv6 host-identifier, the Never Ending Thread: A Summary

[Marcus Goller]

John Jason Brzozowski wrote:
> Frank, et al,
>
> I planned on participating in the 160+ email thread but got tied up.
>
> For what it is worth as someone leading a large IPv6 effort where, as 
> you stated below, MAC addresses are leveraged extensively I can tell 
> you how I handled this issue.
>
> When we specified the use of DHCPv6 in DOCSIS 3.0 we of course 
> leveraged vendor information options.  In these options as you will 
> see if you read the DOCSIS specifications we made provisions to carry 
> the MAC address of the device that adheres to this specification.  
> This is one part of the equation.  I also had to make sure that the 
> necessary back office systems, DHCP for example, account for the 
> presence of these options to support the deployment.
>
> If you wish I would be willing to discuss further offline.
>
> Regards,
>
> John
John,

Interesting point actually, now that you bring it up. When I first 
looked at the DHCPv6 RFC, I thought "Hey! Where did all my options go?". 
If I understand it right, DHCPv6 tries to cover to core functionalities 
only, the rest is left to the vendor information options. The CableLabs 
specifications are a good example, because they might also be 
interesting for people who use a TFTP server as part of the boot process.
Getting client vendors to support an arbitrary vendor information 
option, might be as hard or easy as getting them to support an optional 
extension to the protocol. On the server side it is probably easier to 
agree on a few standard vendor options which get implemented.
The only hard part is to know which vendor information options are 
already out there and are useful, and which need to be created. But it 
is probably the cleanest and intended way of extending the protocol.

Regards,
Marcus