Re: [v6ops] Please review the No IPv4 draft

[Brian Haberman <[e-mail]>]

I purposely picked Owen's message since it is a perfect entry point...


On 4/14/14 3:32 PM, Owen DeLong wrote:
> 
> On Apr 14, 2014, at 10:45 AM, Matthew Petach <[e-mail]>
> wrote:
> 
>> 
>> On Mon, Apr 14, 2014 at 10:26 AM, Nick Hilliard <[e-mail]>
>> wrote: On 14/04/2014 18:23, Matthew Petach wrote:
>>> (which is to say, the potential for abuse here seems kinda high;
>>> are we sure this a good road for us to be traveling down?)
>> 
>> This is no different to any other type of rogue dhcpv4 situation.
>> 
>> Nick
>> 
>> 
>> Correct me if I'm wrong, though; being an ICMP response, rather
>> than a DHCP response would mean DHCP snooping wouldn't be
>> sufficient to stop me from engaging in mischief, where today
>> settings like DHCP snooping and DHCP guard could prevent me from
>> acting as a rogue DHCP server?
>> 
>> I suppose if we extend the concept of DHCP snooping to also include
>> ICMP snooping, that would work.
>> 
>> Thanks!
>> 
>> Matt
>> 
> 
> ICMP was just a suggestion. If you want to put it in DHCP/UDP to
> dodge abuse potential, I have no problem with that. The point is that
> this has no business being encoded in DHCPv6 or RA, it belongs in
> IPv4 somewhere.

Further down the thread, there have been arguments about IPv6 being the
only transport available, hence the need to put this option in
DHCPv6/RAs.  Seems to me the way you keep this option in its own address
family (i.e., carry it in DHCPv4) is to leverage a draft now going
through IETF Last Call (draft-ietf-dhc-dhcpv4-over-dhcpv6).  That draft
specifies how to carry DHCPv4 options in DHCPv6 (due to lack of IPV4
transport).

Regards,
Brian