Re: Confirmation to advance: draft-ietf-6man-ipv6only-flag-05

[Brian Carpenter <[e-mail]>]

Raymond, we are targetting dual stack hosts with an o/s that has extra code
to avoid wasting resource on IPv4. Please re-read the draft.

Regards
    Brian
    (via tiny screen & keyboard)

On Tue, 28 May 2019, 11:07 Raymond Burkholder, <[e-mail]> wrote:

> On 2019-05-27 5:54 p.m., Brian E Carpenter wrote:
> > Then stop telling people to filter Ethertype 0x0800 because you can't
> use RFC2563 in the presence of such filters.
> >> Those are different options, depending on your situation.
> > Well, yes. But if an operator chooses to filter 0x0800 and wishes to
> actively
> > inform hosts that IPv4 is unavailable, only an IPv6 mechanism can do so.
> > So here we are.
> 'only an IPv6 mechanism can do so'  -- is that necessarily true?
>
> This time I did read through
> https://tools.ietf.org/html/draft-ietf-6man-ipv6only-flag-05.  I see
> there is an inclusion of the IPv4 Sunset draft for turning off ipv4 (and
> someone else did point out that the Sunset WG has been itself
> 'sunsetted' I believe).
>
> I also took a look at table called 'EtherType values for some notable
> protocols' at https://en.wikipedia.org/wiki/EtherType
>
> A network link may have many other protocols other than ipv4 or ipv6:
> PTP, MacSec, LLDP, PBB, CFM, RoCE, ATAoE, MPLS, PPPoE, .... not to
> mention STP and the like which is very noisy.  I didn't see a summary in
> the document of how that might be handled.
>
> So I guess this is IPv6-only in the sense of IPv6 with no IPv4.
>
> I can't remember if there was mention of IPv4 only hosts knowing how to
> evaluate an IPv6 flag in an IPv6 protocol, or how vendors
> would/could/want to implement such a thing.
>
> Maybe rather than saying IPv6 only, maybe create another EtherType which
> network stacks would know how to interpret, and it could be used to turn
> off other protocols.  Maybe some networks might want to run IPv4 and
> turn off IPv6 (saying half jokingly).
>
> This independent protocol could be used to turn off other protocols that
> the network stack might see as well.  The flags then become universally
> applicable to any stack that has been designed to interpret xxx-only
> flags.  It would then have long lasting applicability, as protocols come
> and go.
>
> Raymond.
>
>
> --------------------------------------------------------------------
> IETF IPv6 working group mailing list
> [e-mail]
> Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
> --------------------------------------------------------------------
>