On Dec 5, 2011, at 10:58 AM, Noel Chiappa wrote:
From: Bob Hinden <bob(_dot_)hinden(_at_)gmail(_dot_)com>
As far as I can tell, it would only require the CPE router, CGN's, and
routers between the CPE and CGN's to support it. ... I think it's
reasonable for the ISPs who want to deploy this CGN gear to the deal with
upgrading the CPE routers of their customers.
The problem is that for a lot of ISPs, the CPE equipment is owned by the
customers, and comes from a zillion different vendors, and upgrading all of
them is just not feasible.
Sure, for all forms of CPE currently deployed. I assume that there aren't any
CPE deployed behind CGN today. That is, all CPE today work without CGN and any
sort of special addresses, probably use public IPv4 space on the WAN/ISP side.
So a CGN deployment is a new deployment and the ISPs choosing to do this could
make sure that their customers CPE can support class E addresses, upgrade the
CPE firmware, or send them new CPE.
The proposal to use some of the remaining public IPv4 space for this,
IMHO has everyone else incur the costs.
Scenario I: The IETF defines a /10 for this use, to be shared by all ISPs.
Scenario II: The IETF refuses to define a /10 for this use, each ISP goes
out and asks for its own.
Scenario II is incurring less cost on 'everyone else'... how?
Which does lead me to something I've been wondering about, though.
Why don't the ISPs get together, outside the IETF (I so wanted to expand on
this thought, but I had better not), and have one of them - one which is in
an area with an RIR with the most available space - go their RIR and ask for
a /10 for their in-house CGN - and then publish the number and say 'hey,
everyone, here's a /10 that anyone can use for their CGN'? And then, _once
it's already allocated_, they could even publish an I-D/non-standard-track
RFC documenting it, and how to use it...
Ietf mailing list
Ietf mailing list