Date: Fri, 2 Dec 2011 15:20:34 -0800
From: Ted Hardie <ted(_dot_)ietf(_at_)gmail(_dot_)com>
| Big enterprises buy small ones; sometimes at a great rate.
And this itself is a good argument that 1918 space is sufficient (one way or
another), not the reverse. What you have there is two 1918 spaces
(typically, if it isn't, it is irrelevant here) colliding - with just the
same possibilities of collisions as are being mooted here - and yet as you
say, this happens all the time, and I don't see the big enterprises clamoring
for a dedicated address block for their use, so they can avoid clashes when
swallowing smaller ones.
This also raises another problem with the proposal - it assumes that there
are two distinct classes of networks, the carrier ISP networks that get to
use CGNs, and the new network block, and end users, who are supposed to use
1918 space. While I agree in many cases it is easy to categorise users as
one or the other, I believe that there are very many ambiguous cases, ISPs
who believe themselves to be carriers where most of the rest of us would
treat them as end users, and others (like just big enterprises like above
that have swallowed lots of little ones) that are acting more like a big
ISP, even though their only customers are their acquired enterprises.
I still believe that there are just two types of addresses (multicast etc
aside) - routable ones, and non-routable ones, and attempting to set aside
pieces of either of those spaces for one use, as opposed to another, is just
Ietf mailing list