Re: Fwd: Suggestion: charging for IPv4 space

> ... it's a question of costs, direct and indirect.  The cost of
> implementing CIDR was relatively low and its inconvenience level lower
> still, in that there was no flag day or real protocol level
> incompatibility.

i don't understand the distinction you're trying to draw.  folks had to
upgrade all their router images from "could only speak EGP and BGP3" to
"could also speak BGP4".  they had to upgrade every peering session they
had, from BGP3 to BGP4.  in some cases, the router hardware could not be
upgraded (AGS or so) to the new software and had to be replaced.

how is this different from IPv6?  noting that neither has a "flag day"
but both are rolling upgrades that have to touch all nodes of all networks.

is it just that end hosts also have to be upgraded in IPv6, not just routers
as was true for CIDR?

> Also, CIDR was implemented at a time when IP wasn't nearly as
> popular or as mission critical as it is now.

that might be true in retrospect but that's not how it felt at the time.

> ... the ISPs that I dealt with that used NAT hated it thoroughly at all
> levels throughout their businesses, because it hurt their business models
> badly.

agreed.  i just saw internal operations e-mail here that said:

	* Got the Cisco 7970 phones to work ... Several people at HQ are
	using them. If this remains a success I will have others at HQ use
	them. I will also write up instructions on how to configure
	them. Using them behind a NAT appears problematic but I have found
	a way to do so at home. This may not work for everyone so for now I
	think it's best to keep the 7960's for home use.

so i think anyone who thinks there's a steady state solution involving IPv4
and double/triple NAT has to account for these hidden opportunity costs.

> Worse still, enabling ipv6 at the access layer is not going to help anyone
> until real quantities of content also support ipv6 (outside using nat-pt).
> How can Joe Bloggs ISP or John Doe Telco request that Google, Yahoo and
> Microsoft put quad-A records into their DNS?  It's difficult to imagine
> turning off ipv4 any time in the forseeable future.

google, yahoo, and microsoft will put quad-A records in their DNS as soon as
it makes business sense (which is a cost:benefit question) for them to do so.
therefore unless you think all new traffic growth can be made NAT-compatible,
it's safe to assume that these quad-A RRs will appear at or near the time of
IPv4 runout.

> Sorry that this sounds like beer-chat, but we are attempting to quantify
> pain here. Everyone's pain threshold is different.  Some companies will
> be innovators, and some laggards.  Some will stick to their guns that
> using ipv4 is their inalienable right and will never implement dual stack
> systems.  Others will use nat-pt from the time that it becomes feasible
> to do so.

these are excellent observations and i agree with your summary here.

> Nick

paul
_______________________________________________
global-v6 mailing list
[email protected]
http://mailman.apnic.net/mailman/listinfo/global-v6