Re: [**EXTERNAL**] Address length needs

["Joel M. Halpern" <[e-mail]>]

Fair enough.  Let's not use the phrase "race to the bottom"

We can observe that some operators have indicated in reputable surveys 
that they would like to allocate /126 to customers.

We know, from both experience and analysis, that any particular number 
is hard to defend.  We could say the longest assignment to a UE / Node / 
... had to be /96.  And we would get arguments about why it needed to be 
longer.  We agreed during the design of IPv6, on the /64 in part so that 
we would not have to repeat the argument eveyr time we turned around.

there is a fair question about how an operator who is not using DHCPv6 
for address assignment (whether mobile or fixed) can use RA to give out 
a prefix to a piece of equipment that can be used for suballocation. 
The premise in this discussion is that such an allocation would need to 
be longer than /64.  I do not understand why.  If we are going to etend 
RA to support this (a request that seems reasonable) then why not do the 
extension in a way that respects the /64.

It would seem to be sensible to look at solutions that do not violate 
MUSTs in existing specs before assuming solutions that change the MUST 
are necessary.

Yours,
Joel

PS: The airplane case is sufficiently different that I find it hard to 
include it in the discussion.  In part because there is a LOT of design 
that we are not free to alter in order to ask "is that the right 
answer".  It is absolutely fair and necessary for folks to do designs 
for their own problem space.  But those designs should respect the 
limits that have been in place for more than 20 years.  Or be prepared 
to have actual discussions of design choices.  (Tony and Fred mean ve3ry 
well, and they are doing a good job of describing the constraints ICAO 
has accepted.  But they can't change those constraints or make commit to 
make changes.  Heck, it is not clear they even agree with each other.)

On 11/8/2020 11:56 AM, Nick Hilliard wrote:
> STARK, BARBARA H wrote on 08/11/2020 15:07:
>> I agree. The "race to the bottom" argument is a 100% FUD argument.
> 
> It's more than just FUD. The term "race to the bottom" is a politically 
> charged phrase which is being used to shut down and obscure legitimate 
> discussion about the complex issue of subnet size management on 
> production networks.  Loaded language of this form has no place in an 
> SDO working group.
> 
> Nick
> 
> --------------------------------------------------------------------
> IETF IPv6 working group mailing list
> [e-mail]
> Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
> --------------------------------------------------------------------