Re: Updated IID length text

Le 19/01/2017 à 20:26, Brian E Carpenter a écrit :
> On 19/01/2017 22:26, Alexandre Petrescu wrote: ...
>> I do not agree with the part that says "IIDs are required be 64bit
>> long".
>
> But they are, today. In the move to full Standard, we cannot actually
> change that; we can only clarify the wording.
>
>> If IIDs are required to be 64bit long (even if only in the
>> 000-prefixed space) then the network can not grow at the edges, at
>> least not with SLAAC.
>
> That would be true if /64 prefixes were scarce. They are not.

The mobile network operators today make it look as if the /64
prefixes were scarce.

Maybe they dont know these /64 are not scarce, or maybe they dont care
about scarcity (just as it's not IPv4 scarcity making them deploy IPv6).
  The end result is the same: end users get each a /64, i.e. one single
subnet.

Spec-wise the IID len cant change, the Ethernet IID len cant change, the
DHCPv6 PD is not deployed, the operators only give a /64 to end user...
one is free to pick the reason, and again the end result is the same: no
growth at the edge.

The archi document is free to recognize reality.  If so, it should
recognize it so in a paragraph: no growth at the edges.

Alex

> But that's why the proposed text is limited to currently allocated
> unicast space - if we ever need to change this, it will be a new
> tranche of unicast space. In any case, this is nothing to do with
> progressing to full Standard.
>
> Brian
>
> --------------------------------------------------------------------
>  IETF IPv6 working group mailing list ipv6@ietf.org Administrative
> Requests: https://www.ietf.org/mailman/listinfo/ipv6
> --------------------------------------------------------------------
>