Re: IPv4 traffic on "ietf-v6ONLY"

Do we provide a DHCPv6 server and set the O bit in the RA to provide the DNS server addresses?  Manual config shouldn’t be required if we do that unless your OS is broken. I’m not on site to check. 

-- 
Mark Andrews

> On 16 Nov 2017, at 00:29, Alexandre Petrescu <alexandre.petrescu@gmail.com> wrote:
> 
> 
> 
>> Le 15/11/2017 à 12:30, Warren Kumari a écrit :
>>> On Wed, Nov 15, 2017 at 7:02 PM, Rajiv Asati (rajiva) <rajiva@cisco.com> wrote:
>>> 
>>> Of course, a DHCPv4 option conflicts with the goal of some people to have
>>> no IPv4 related infrastructure at all.
>>> 
>>> 
>>>  Perhaps, define a DHCPv6 option to convey v6-only, for which the client
>>> interpretation should be to suppress v4.
>> Can someone remind me what problem we are trying to solve?
> 
> To me, the problem comes mainly from the impossibility to say "IPv6-only".
> 
> It's not just a terminology problem.  There are bad side-effects of abusing the term "IPv6-only".  Some people claim that IPv6 is there already, because there are "IPv6-only" networks.  But there are no "IPv6-only" networks at all.  (except in some very rare cases that are not visible at IETF nor at mobile network operators.)
> 
> Another problem can be related to that wireless network efficiency (too many useless IPv4 packets on the air of "ietf-v6ONLY").
> 
> The 'noise' problem is symmetric - valid in IPv4 too.  The person at the IETF ticketing system tells me I have to go meet them to manually remove the DNS Server IPv6 address that the "ietf" SSID has put on my Windows and that stays there forever.  That DNS IPv6 address makes noise on my corporate IPv4-only network.  I suppose the person does not want me to go there at each IETF for that manual operation.  That's a speculation of a problem too.  Maybe he has some magic means (a protocol?) that makes that the DNS IPv6 address has a lifetime, or so.
> 
> Alex
> 
>> W
>>> 
>>> Although this will be at the cross road with allowing client’s wishes to use
>>> v4 LL for whatever useless/useful traffic, it would be a reasonable
>>> deployment policy to enforce.
>>> 
>>> We have to remember that allowing this would continue to hurt WLAN (not
>>> wired LAN) for obvious reasons - radio bandwidth.
>>> 
>>> During one of ciscolive conferences, i recall having observed lots of
>>> useless v4 traffic (e.g. discovery) on WLAN.  Andrew (cced) may remember
>>> more details.
>>> 
>>> Cheers,
>>> Rajiv
>>> 
>>> On Nov 15, 2017, at 5:48 AM, Philip Homburg <pch-ipv6-ietf-4@u-1.phicoh.com>
>>> wrote:
>>> 
>>> Does that return us to the question of how to tell hosts that IPv4
>>> 
>>> doesnt live here, and to stop trying?
>>> 
>>> 
>>> The safest option to do that is a DHCPv4 option that says 'no IPv4 service
>>> here, go away'.
>>> 
>>> Any network that has IPv4 production traffic already has to protect against
>>> rogue DHCPv4 servers.
>>> 
>>> Of course, a DHCPv4 option conflicts with the goal of some people to have
>>> no IPv4 related infrastructure at all.
>>> 
>>> --------------------------------------------------------------------
>>> IETF IPv6 working group mailing list
>>> ipv6@ietf.org
>>> Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
>>> --------------------------------------------------------------------
>>> 
>>> 
>>> --------------------------------------------------------------------
>>> IETF IPv6 working group mailing list
>>> ipv6@ietf.org
>>> Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
>>> --------------------------------------------------------------------
>>> 
> 
> --------------------------------------------------------------------
> IETF IPv6 working group mailing list
> ipv6@ietf.org
> Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
> --------------------------------------------------------------------