[arin-ppml] 2008-6: Emergency Transfer Policy for IPv4 Addresses

[Tom Vest]

On Oct 9, 2008, at 12:39 PM, Milton L Mueller wrote:

>
>> On Oct 6, 2008, at 9:27 AM, Randy Bush wrote:
>>
>>> and this is not a problem if you get the space from an rir, only if
>>> you buy it on an open market.  cool!
>>>
>>> randy
>>
>> That's right Randy. To recap from last week, what makes routing keep
>> working under the current paradigm?
>>
>> 1. CIDR  -- which provides the basic tools.
>
> Which still exists if there are transfers...

Hi Milton,

Perhaps you misunderstood something(s).
"Open market" means that IPv4 address resources obtained for use after  
the exhaustion of the RIR-administered reserves can come from a wide  
variety of sources. That's the *opposite* of "top-level aggregation."  
CIDR was designed to manage the task of routing system capacity  
conservation using the inputs that top-level aggregation permitted.  
CIDR does indeed still exist with or without those inputs, but without  
them it's of no further use as a check on the ever-growing demand for  
routing system capacity.

Maybe individual IPv4 dealers will attempt to optimize their future  
IPv4 sales to maximize aggregation potential -- i.e., public benefit  
-- rather than profit/private benefit, but widespread assumptions  
about the inevitability of IPv4 sales regardless of rules or policies  
doesn't exactly support that assumption. Even if they did try, and  
even succeeded, the growth trend under a single centrally administered  
"top-level aggregation" is already alarming to some operator. *That*  
was the basic gist of Randy's oft-repeated critique of the RIR system.  
Multiply that trend by a thousand, or a hundred, or maybe just a  
handful of new top-level address resource suppliers, each with strong  
incentives may often conflict with aggregation, and you can see how  
that demand trend might change a bit.

Buggy whips still exist too, but I find that my car's accelerator  
pedal works better.

>> 2. Top-level aggregation -- which the RIR community-system provided,
>> and kept flexible over time as technology improved and RIR community
>> practices changed.
>
> Which still exists if there are transfers...
>
>> 3. Filtering -- which was/is only commercially feasible because of  
>> the
>> top-level aggregation made possible by the RIR community-system.
>
> Which still exists if there are transfers...

This one you clearly misunderstood.
Sure, large service providers can filter long prefixes, but the only  
way they can do that without voluntarily losing visibility to/from  
some parts of the Internet is if all of those long prefixes are  
"covered" by  larger routing table entries -- which is what the  
combination of top-level aggregation plus CIDR makes possible. Without  
that, they can still use filtering to protect their own routing  
subsystem investments, but only at the expense of foregoing new  
customers that arrive with nothing but a small quantity of purchased  
IPv4. In addition, as more growth is accommodated by such small  
prefixes, a filtering service provider will not only miss out on new  
business, they'll probably start lose some of the existing customers,  
who would also be losing visibility to everything new that's coming  
online. On top of that, they'll also be pissing off their peers who do  
not filter, because the filterers will be indirectly degrading the  
connectivity of the non-filterers' new resource transfer recipient  
customers.

The best way to handle this will be a routing cartel of some kind, at  
least until the DOJ steps in.
After that, the best way to handle it will be somebody else's problem.

>> 4. Open entry for new routing service providers, which the arms- 
>> length
>
> Which becomes more limited due to ipv4 scarcity with or without
> transfers

More limited, asymptotically approaching absolutely impossible,  
forever -- that's the future that a self-regulating, privatized IPv4- 
based Internet promises.
I admit that it's in an improbable one, but improbable only because  
the "self-regulating" aspect will be the first to to go.

Frankly, given your long advocacy for expanding the supply of TLDs,  
I'm a little surprised that you've chosen to support artificial  
scarcity when it comes to number resources. When confronted with a  
hard choice between the means (competition, the profit motive, the  
prospect of "cornering the market" once and for all) and the  
hypothetical ends of market-based allocation regimes (e.g., relative  
abundance, the prospect of winners as well as losers being better off  
than under any other arrangement), you've made your own preferences  
abundantly clear. And so have I.

Now the community (or of you prefer, the "community") will have to  
decide.

TV