[ppml] PIv6 for legacy holders (/w RSA + efficient use)

[David Conrad]

Paul,

On Jul 30, 2007, at 9:51 PM, Paul Vixie wrote:
> to the extent that you're talking as if none of us has any choice  
> as to the
> outcome, the word i've been using, "fatalistic", is dictionary  
> perfect.

Individually, we all have a choice. Collectively, Leo was right:  
http://en.wikipedia.org/wiki/Prisoner's_dilemma

> to say that human nature preordains a period of scarcity before the  
> new kind
> of IP addresses start being relevant, goes beyond me, at this stage.

I guess I have less faith in companies to see beyond the end of this  
quarter than you.

> to say that a market would lead to efficient utilization of either  
> routing
> table slots (vs. subdivision) or space (vs. a futures market), and  
> will serve
> humanity as well as a market in pork bellies or google shares, goes  
> beyond
> me, at this stage.

I have said on several occasions that increased address utilization  
that will come from an address market will result in a flood of long  
prefixes and that I worry about this.  However, the IETF backed away  
from the output of the RAWS workshop and I have been told that YFRV  
(both colors) have stood up at NANOG (etc.) and claimed 2M prefixes  
today, 10M soon with no change in hardware, and that they're not  
worried.  Who am I to contract these learned groups?

> a lot of people have said that a market is inevitable and a lot
> of people have said that a market would be efficient.

A market in address space already exists.  As to whether it will be  
efficient, sure: for some value of "efficient".  Utilization  
efficiency will increase.  Routing efficiency will decrease.  Money  
will move from one place to another.  So it goes.

> instead, here's a rim shot: what's your preferred outcome?

That IPv6 provides its own incentive to migrate.

I once had a dream: IPv6 would be able to provide something that  
isn't possible in IPv4, namely a simple mechanism to allow for (non- 
BGP) multi-homing, mobility without the complexity of MobileIP, and  
didn't require people to renumber when they changed providers.  The  
LOC/ID split, separating the end point identifier from the locator  
could have been done in IPv6 where it was essentially impossible in  
IPv4.

There is still a small glimmer of hope.  However, failing that, I  
figure it's going to get really ugly and I have a fear governments  
are going to get involved in the end.  Hence, my desire to extend the  
lifetime of the free pool via proposals such as "Soft Landing"...

Rgds,
-drc