[ppml] FW: No transfer policies are needed
stephen at sprunk.org
Mon Apr 21 16:40:08 EDT 2008
Thus spake <michael.dillon at bt.com>
> When the IPv4 free address pool is exhausted, i.e. fully allocated to
> organizations who NEED those addresses to USE them in the network,
> it seems highly unlikely that there will be any significant number
> of UNUSED IPv4 addresses anywhere. The fundamental reason that the
> free pool is exhausted is that the network has grown so large that
> all those addresses are needed for operations. People like to point
> to some of the early /8 allocations as wasteful, but there is no
> evidence to support this. Many organizations who were not using their
> original /8 or /16 allocations, have returned them to the free pool.
> Chances are that most of the remaining allocations actually are in
> use and that those organizations have designed a network which does
> not use Port-NAT because they built their networks before Port-NAT
I know of a large number of enterprises that have one or more legacy Class B
networks that are not advertised at all; they're used on the private side of
a NAT or proxy setup. Such orgs could easily switch most of their hosts to
10/8 within a few days; the biggest obstacle is inertia and change control
processes, not technical limitations, and could be easily overcome by a
directive from a CFO looking at potential revenue.
> All the people who have actually crunched some numbers over recovery
> of /8s and /16s, come up with just a few months of extra time until
> IPv4 exhaustion. This is the pool of addresses that would theoretically
> be transferred under some form of transfer policy. It seems rather silly
> to put so much effort into something with such small addressing impact.
I don't believe anyone has run the scenarios you claim because the necessary
data simply isn't available.
There _have_ been studies on how much time we'd get by reclaiming legacy
space that is _currently unused_, and that usually comes out to a few
months, perhaps a year. However, AFAIK, none of those studies considers
fractions of blocks _currently used_ that would be freed up if there were a
financial incentive to do so. Nobody has that data.
For instance, my company has a legacy Class B network that is "in use".
However, it's entirely possible that the finance folks would decide that,
for a sufficient amount of money, we'd renumber to 10/8 and sell the Class
B, or perhaps renumber into a /24 subnet (for our public servers) and sell
the rest. I'm not aware of anyone having offered us money, so I have no
idea what our CFO's reaction would be.
> Paradoxically, in addition to harming the people who are
> unable to buy addresses,
They are no more harmed than they would be if exhaustion hit and there were
no market. This is a red herring.
> the transfer policy HARMS the organizations who succeed in
> buying addresses because they lose large sums of money which
> reduces their ability to move to IPv6.
It is not up to you to tell other organizations how to best allocate their
funds. If some folks decide that paying for IPv4 space is better _for them_
than migrating to IPv6 (if that's even possible), so be it. I'll also point
out that for every party "harmed" in your view this way, there is another
organization _receiving_ money to help fund their migration.
>> 2. It appears to reward large legacy holders who squatted on
>> resources that others in the industry need. ARIN's actions
>> in supporting a transfer policy that recognizes if not
>> actively accounts for the 'selling' of address resources
>> seems to sanction this activity.
> To date, I believe only one AC or BoT member has disclosed their
> financial interests in organizations holding legacy allocations.
> This places suspicion on everything that ARIN does. Every single
> AC and BoT member should make this disclosure and it should be
> placed on the ARIN website where we can easily see who, if anyone,
> stands to benefit personally from a transfer policy.
Disclosures are only required when someone has a conflict of interest; lack
of disclosures does not imply people are hiding things. Unless you're
accusing AC/BoT members of ethics violations, one should assume that those
who haven't made a disclosure have nothing to disclose.
Furthermore, neither the AC nor BoT actually makes a decision on policy --
the community does. The BoT rubber-stamps the AC's decisions unless there
is evidence that the IRPEP was not followed (which is rare). The AC gauges
community consensus, regardless of the members' individual views. As others
have pointed out, members of both bodies have recused themselves (as shown
in the minutes of their meetings) to avoid even the appearance of
Stephen Sprunk "God does not play dice." --Albert Einstein
CCIE #3723 "God is an inveterate gambler, and He throws the
K5SSS dice at every possible opportunity." --Stephen Hawking
More information about the ARIN-PPML