Re: [nvo3] Number assignments

[[e-mail] (Dale R. Worley)]

Alia Atlas <[e-mail]> writes:
> As a matter of courtesy to the IEEE, with whom the IETF has a very
> good relationship, I see absolutely no reason that we would not
> respect their rules and suggestions for managing their own code spaces
> - just as we expect the same courtesy for our IANA managed registries.
>
> If there's actual interest in going down this path for implementation
> & deployment, I would strongly recommend using the sub-types approach
> that Donald suggested.  That is the approach that the IEEE recommends.

I'm not sure how you're parsing the situation.  But of course, we could
not obtain Ethertype(s) unless IEEE assigned it(them), so contravening
IEEE's practices isn't possible, in the strict sense.

At the moment, I'm envisioning that we would redefine the Geneve
"protocol type" field so that values of 256 and larger are to be
interpreted as Ethertypes and values of 255 and less are to be
interpreted as IP protocol numbers.  Clearly, this is overloading the
field, but conveniently, 802.3 forbids Ethertypes less than x0600 or so,
so there is no ambiguity.

>From my point of view, this isn't an abuse of Ethertypes or the IEEE's
policies, it's taking advantage of stated properties in the standard.
But perhaps the IEEE wouldn't look at it that way.

Dale