Bill Manning <bmanning(_at_)ISI(_dot_)EDU> writes:
There is near zero value in the number/address and
very real value in the routing slot. Perhaps it is best
to simply have ebone route filter on the /16 boundaries
to drive home your point. (being cranky this morning)
I utterly reject your ostrich-like position on this matter.
I would be extremely happy if I could make a money-based
or (better still) capacity-based offer to one of the R&E
networks or institutions which retain very short prefixes
(historically known as Class A and Class B networks),
without being prevented from engaging in such a private
transaction by the collusive behaviour of IANA and the registries.
Likewise, I think that government agencies in various East
and South Asian countries, and perhaps various Asian ISPs
or next-generation mobile telephony organizations
would be extremely happy to bid for a few historical Class As
now being under-used by the current registrants, rather
than being told "no" or "first demonstrate usage" by APNIC.
It has been made clear in the past that any transfer of
addresses will be reviewed by a registry, and that if the
ultimate recipient of transferred address space wants
more address space from the registry, they must comply
with the ordinary "growth & design" rules.
I hear anecdotally that the threat of a withholding of new
allocations to the selling party has also been made in the past.
This is a system which enforces a "one-seller" (the IANA),
"one buyer" (one may return addresses to IANA only) model,
which flies in the face of free markets, and perversely
imposes costs upon consumers.
Although I am happy that there are people trying to
conserve IPv4 addresses and also encourage sensible
routing announcements by providing not less than a
sizeable aggregatable range to qualified buyers, the
qualification process is tricky and gets trickier as
one's business grows.
There is a VERY real cost -- most notably in terms of time --
to using the "growth & design" scheme for acquiring more addresses
than an initial allocation. This has, in fact, slowed the
deployment of independently-routed subnets owned and
operated by a single organization. This slowness could
have been avoided if a market for IP addresses existed,
and I can assure you that a fairly sizable amount of money
would have been spent to speed up the process of acquiring
a handful of relatively long prefixes. I am also aware
of anecdotal reports of organizations who had to suspend
turning up newly-acquired customers because they could not
quickly acquire new addresses from the monopoly vendors:
the local registries and IANA.
In my opinion, following the PIARA work, the appropriate
thing for IANA to do is to spin off its IPv4 address
allocation function. We will call this spin-off IANA-I.
IANA-I should then proceed auction off the *ENTIRE*
_not-yet-allocated_ address space, being very clear that
what is sold at auction is merely an exclusive
registration of a range of IP addresses in an
IANA-maintained and publically-accessible document, the
right to make future changes to the registration, and the
right to transfer future registration change rights to
The IANA-I or its agents could certainly charge a small fee
for processing such changes from persons duly authorized
by the registrant the IANA knows, however it should not
have the power to refuse any transfers of title.
I would furthermore like to see the ENTIRE unallocated
IPv6 global unicast space auctioned off in a similar manner,
abandoning the anti-market "one-seller" model put forward
in RFC 2450 section 5.0, using the monopoly tariff put
forward in section 5.2 ibidem.
That ISPs probably cannot be compelled to consider the
IANA-I registration document at all, in whole or in part,
when configuring their networks' routing policies, should
be declared by the IANA and its auctioneer agent, much as
the registries note this now when making allocations under
the current "one-seller" rules. Whether there is value in
such a risk-bearing instrument,however, MUST be determined
by buyers, not by IANA, IANA-I, or ivory-tower academics.
There are certainly MUCH riskier instruments traded
regularly as assets on exchange markets throughout the
financial world. Moreover, the IPv4 black market that
DOES exist, as noted by David Conrad, argues strongly in
favour of testing the "white market" in a sensible fashion.
Finally, a small initial registration fee by IANA-I could
allow currently allocated address space could be noted
in the IANA-I registration document, thus normalizing the
"deed" to the range of addresses, likely making it easier
to undertake a transfer.