[arin-ppml] IPv4 Transfer Policy Change to Keep Whois Accurate

[Owen DeLong]

On May 20, 2011, at 6:20 AM, Mike Burns wrote:

> Good Morning Owen,
> 
>>> Like those unfortunate souls who will find themselves behind CGN will likely enrich their ISP, who can sell as many addresses into the market as he can without losing too many customers who demand >>non-CGN access.
> 
>> So you are saying that your proposal will increase the deployment of LSN... Yet another reason to oppose.
> 
> No, I am saying the deployment of LSN is likely to happen, and when it does an uncertain number of ip addresses may be freed up to supply the transfer market.
> 

Either your proposal makes a difference here, or, your above statement isn't relevant to the discussion. If your
proposal makes a difference in this regard (encouraging LSN), that is a negative IMHO.

> 
>>> Or even the deployment of plain old NAT could free up addresses which are currnently being  advertised now for sale later.
>> 
> 
>> So you are saying that your proposal will increase the deployment of NAT... Yet another reason to oppose.
> 
> I am saying that your "proof" of the size of the transfer pool is not valid because you assume 100% efficiency in routed space.

I actually don't assume that at all, but, that's OK. There may be other assumptions built in. I would argue that increasing
NAT does not increase the efficiency of address utilization, but, merely increases the number of users on a single
address while decreasing the utility of those addresses. The net efficiency remains roughly the same if one considers
the utility factor as part of the efficiency equation.

So, again, either your proposal will make a difference here (encouraging NAT) or it won't. If it doesn't, there's
not a lot of relevance to your statement.

> 
> 
>>> The pool of buyers is known to be monotonically increasing and ARIN's average issuance
>>> of ~190,000 IPv4 /24s each year by ARIN would indicate an annual market demand for 25%
>>> of the maximum possible pool from which to develop such a market (roughly 12 /8s)
>> 
>>> Therefore, even if we assume that all possible addresses will come on the market with
>>> your policy, the supply will be gone in 4 years or less. Any churn beyond that would leave
>>> a need-gap behind unless it is the result of successful abandonment of IPv4. At the point
>>> where organizations can begin successful abandonment of IPv4 altogether, the market
>>> will invariably collapse anyway.
>> 
>> And you believe this and still think there is a danger from speculators? When the market is at most 4 years in duration and then subject to collapse due to IPv6?
>> 
> 
>> Yes. I think that if you open it up to speculation, the duration becomes much closer to 4 minutes
>> than 4 years.
> 
>> Owen
> 
> Owen, any single entity is limited to a /12. I trust the ARIN staff to be able to determine whether the entity is a sham company, and even by the size of your limited pool, a /12 can't corner the market. So maybe we can at least temper the fear that a speculator would enter into a 4 minute market.
> 

It doesn't have to be a sham company, it can simply be a subsidiary. For example, last year, a certain
cable provider got a /9. They would have no problem whatsoever creating 8 regional operating
companies each of whom needed a /12 in my estimation.

The /12 limit is both absurd and utterly useless in having the effect you claim. All it would really
do is penalize large providers (note, I'm seeking fairness, not penalties for either side) and
increase disaggregation.

Owen