[ppml] alternative realities (was PIv6 for legacy holders (/wRSA + efficient use))

[michael.dillon at bt.com]

> On Jul 30, 2007, at 7:22 PM, Paul Vixie wrote:
> > i'd like to know what makes this alternative the most likely.
> 
> Roughly, the way I figure it, given:
> 
> a) there is a demand for that resource (say, people who want 
> to connect to the Internet)
> b) there is a limited supply of a resource (say, for argument, unused
> IPv4 addresses)
> c) there are entities who have more of the resource than they need

If so, then they are not in compliance with global policies on IPv4
allocation. The only entities that I know of who are in this postion are
so-called legacy holders such as MIT. But I do not believe that entities
like MIT have a right to monetize their IPv4 addresses when the vast
majority of industry is playing by the rules and has no surplus.

> d) there are entities who are willing to exchange value for 
> that resource
> e) "traditional" mechanisms to obtain the resource will not 
> work (no more free pool)
> 
> Then you have a breeding ground for a market (color isn't 
> important here).

Maybe you are right but that is not the necessary conditions to SUSTAIN
a market. If you can't sustain a market in IPv4 addresses through
regular churn i.e. address blocks changing hands, then you don't have a
real market. Instead you have a flash in the pan where a few private
deals get made, and then the IPv4 free pool is utterly exhausted.
Possibly a few of the private deals were purchases by speculators who
now have a monopoly on available addresses and charge an inflated rate
for the last few blocks.

You simply cannot sustain a market under those conditions. In financial
terms, there is not enough liquidity. To have liquidity you either need
a large number of potential sellers or you need a minimum volume of
transactions. Check various financial market liquidity rules to see how
this is done.

If you have insufficient liquidity then you will not have fair prices
and that means you also do not have stability of predictability. Without
stability or predictability, companies will refuse to participate in the
market, further reducing liquidity. Given the nature of these problems,
and the fundamental shortage of addresses to trade, I really cannot see
a market forming, even if the legal issues surrounding "property" and
"ownership" could be resolved.

The only thing that will convince me that a market is possible would be
an official public statement from the U.S. Department of Commerce
stating that it supports the concept of a market in IPv4 addresses.

> then the vast majority of users have no business incentive to deploy
> IPv6 and hence no reason to ask your ISP for IPv6 service.

Running out of IPv4 addresses and putting the brakes on network growth,
are reason enough for most IP network operators to deploy IPv6.
Customers will be forced to go along because they also will be unable to
grow their internal networks using globally unique addresses. I fully
expect ISPs to introduce IPv6 with price differentials between IPv6 and
IPv4 services. If a company absolutely must have IPv4 connectivity with
globally unique addresses, then they will have to pay a higher price.
Smaller, more innovative and agile companies will take the cheaper IPv6
option.

IPv4 exhaustion is a business incentive in and of itself. It is a fact,
like earthquakes in California and hurricanes in Florida. Businesses
must deal with it and spend the money to avoid catastrophe, or when the
wall hits, they will suffer.

> (4) will probably change (call me an optimist), but I argue 
> there will be significant content that is not available via 
> IPv6 for the foreseeable future.

If I were you, I would check with the developers of Squid to see if it
can function on a server with two  network interfaces, one running IPv4
connected to the Internet and one running IPv6 connected to the LAN. I
suspect that if it is not already fully functional, then someone will
discover that it requires only two or three days of development and
testing. You know the effect that "running code" has, don't you?

--Michael Dillon