Re: A 3rd try at a proposal for draft-ietf-6man-rfc4291bis-07

[Lorenzo Colitti <[e-mail]>]

On Tue, Mar 7, 2017 at 12:35 PM, Fred Baker <[e-mail]>
wrote:

> Of a network chooses to do something different in the absence of that
> limitation, I understand the Internet Architecture to give it that latitude
> and the authority to impose an address allocation mechanism and policy.
>

The architecture currently does not give it that latitude, because RFC4291
makes a blanket statement that ("most") IIDs are 64 bits long.


> The argument I have heard for serial lines is that, with high probability,
> if there are more than 2^1 addresses on a link, selecting a random address
> for one's peer has a high confidence of selecting an address the peer isn't
> using, resulting in an ICMP error instead of a response.


I'm not sure I understand this problem statement. Are you saying that for a
link that only has two peers on it, it is convenient to use a subnet that
is /127, because you don't have to know the peer address? That's true, but
it only applies to /127 and not any other prefix size, right?


> That, assuming I understand it, is the fundamental argument behind /127.
> Reduced to simpler terms - if the environment makes assumptions, the
> addressing should make the same assumptions.
>

Isn't that still a philosophical argument that applies to the current
solution, though?