[arin-ppml] ARIN Response to AFRINIC on Policy compatibility

Owen DeLong owen at delong.com
Mon Jan 23 18:55:15 EST 2017

> On Jan 23, 2017, at 08:41 , Mike Burns <mike at iptrading.com> wrote:
> Hi Owen,
> May I point out that despite reciprocity with APNIC, almost no addresses
> have flowed from APNIC to ARIN?  I think less than a /17 in aggregate since
> the first interregional transfer in 2012.

I’m well aware of this. What I don’t know is why you would consider that a relevant point.

My reasons for wanting to preserve the requirement of reciprocity are much more a matter of
principle and along the lines of Joe’s concerns about balkanization with NIRs such as CNNIC’s
abhorrent policy of “addresses check in but they don’t check out and all addresses used in CN
must be in CNNIC”.

> You are correct in your expectation that actual transfers aren't
> symmetrical, because they respond to market forces.

Right, so I see no need to abandon the requirement of reciprocity, nor do I see a reason for
the other RIRs to avoid providing a reciprocal policy if they want to play on the worldwide
transfer stage.

> So we are saying to the address-poor regions of the globe that we refuse to
> send our addresses to them because we are standing on principal, even if
> that principal has little real effect.

I suppose if you want to put it that way, modulo the misuse of principal vs. principle,
yes, that’s similar to what I am saying. What I would say is that rather, we are not
buckling to pressure to create harmful asymmetric policies simply because the regions that
want us to do so happen to be some of the most address-poor regions in existence.

Frankly, I’m personally tempted at this point to throw my hands up and let IPv4 become an even
more obvious cesspool of disaggregation and watch the melting of the IPv4 routing table
become the driving force for urgent IPv6 adoption. However, that’s not in line with the
responsibilities I have to the community in my elected position of trust or the principals
of good stewardship.

So, for now, I will continue to make the principled argument against asymmetric transfer
policies because I believe the precedent they create is harmful. We already have two rather
awful examples of where such policies are causing real problems (CNNIC and VNNIC) and I
do not want to do anything to encourage expansion of this precedent.

> The communities in these regions can regard the history of allocations and
> reasonably come to the conclusion that they have been shortchanged.  Thus
> they might feel less inclined to stand on that principal of reciprocal
> trade, which I agree is the best policy.

They can have all the IPv6 they want, so the other way they could look at this is that they’ve
been allowed to avoid the whole overhead of additional IPv4 deployment in favor of leapfrogging
the technology to IPv6.

However, neither of these perspectives is completely accurate or useful in the discussion at hand.

> I talked to some LACNIC members who expressed an unusual fear to me, a fear
> based on the difference in economic realities in the Southern versus the
> Northern Hemisphere in the Americas. The fear was that poorer LACNIC members
> would decide to re-engineer their networks to take maximum advantage of
> CGNAT for the purposes of selling their addresses, and the fear is that
> these sales will be to the richer regions of the world, resulting in outflow
> and degraded local Internet.  Thus a potential danger is present in some
> minds which a unidirectional policy would obviate.

Except that it won’t really do anything to change that. All it will do is limit the
sales of those addresses to the northern companies that have operations (or create the
appearance of operations) in those southern regions.

> This reminds me of when ARIN also required reciprocity in needs-testing. In
> my opinion, the ARIN community used its historical imbalance of address
> allocations to cudgel the APNIC community into changing their policies to
> meet ARIN's demands, for a likewise minimal effect, in order to stand on the
> principal of needs-testing.  At least needs-testing had historical
> antecedent in the free-pool allocation policies. However there never was any
> history of reciprocity in inter-regional transfer policy, that seems to be a
> new principal and not one worth engendering conflict with LACNIC and

To the best of my knowledge the compatible needs-basis requirement has not been
amended since it was put in place as you describe. Yes, APNIC chose to modify
their transfer policy in support of being able to work with ARIN on inter-RIR
transfers. If you want to call that a cudgel, then that is your right. I call it
a great example of how making a principled argument and good policy can actually
create positive change.

> As far as this policy opening the door or setting a dangerous precedent, may
> I point out that this one-way policy has been operational for years
> regarding certain Asian NIRs, and the precedent has not proved dangerous.

Actually, it has proved quite harmful IMHO. I’m not at liberty to disclose the
specifics of any of them, but I know of several situations where organizations
wanting to do business within those jurisdictions have been forced to obtain and
transfer addresses out of their control at their expense with no hope of ever
regaining control of those addresses even if they shut down their operations in
the jurisdictions in question as a result.

I suggest you discuss the matter with a few multinational organizations that have
any sort of significant operational footprint within the jurisdictions in question
before you declare such policies harmless.


> Regards,
> Mike
> -----Original Message-----
> From: Owen DeLong [mailto:owen at delong.com] 
> Sent: Friday, January 20, 2017 8:29 PM
> To: Mike Burns <mike at iptrading.com>
> Cc: Job Snijders <job at ntt.net>; Scott Leibrand <scottleibrand at gmail.com>;
> ARIN-PPML List <arin-ppml at arin.net>
> Subject: Re: [arin-ppml] ARIN Response to AFRINIC on Policy compatibility
> The reciprocity requirement merely requires that the policies ALLOW
> transfers in both directions.
> I do not believe that allowing transfers to an RIR which will not allow
> transfers out is reasonable or prudent and this belief has nothing to do
> with maintenance or protection of a free pool. If we will allow transfers
> between RIRs, then the policies by which they are allowed should be fair,
> balanced, and symmetrical. This does not mean that I expect the ratio of
> actual transfers to be balanced or symmetrical, merely that the policies
> under which they are conducted should be.
> Owen
>> On Jan 20, 2017, at 09:48 , Mike Burns <mike at iptrading.com> wrote:
>> I forget where the original numbers came from, but with a total of 
>> 130, obviously many /8s are missing.
>> Probably this count is not considering legacy space, most of which is 
>> North American.
>> Including those legacy addresses, the supply for much of the transfer 
>> market, the ratios are much more in ARIN's favor.
>> Regards,
>> Mike
>> -----Original Message-----
>> From: ARIN-PPML [mailto:arin-ppml-bounces at arin.net] On Behalf Of Job 
>> Snijders
>> Sent: Friday, January 20, 2017 7:23 AM
>> To: Scott Leibrand <scottleibrand at gmail.com>
>> Cc: ARIN-PPML List <arin-ppml at arin.net>
>> Subject: Re: [arin-ppml] ARIN Response to AFRINIC on Policy 
>> compatibility
>> On Thu, Jan 19, 2017 at 05:40:55PM -0800, Scott Leibrand wrote:
>>> Why is average /8s per continent the right metric there?  Wouldn't
>>> IPv4 addresses per capita be more like what we're looking for?  I 
>>> haven't run the numbers, but I suspect the ARIN region is higher than 
>>> all four of the other RIRs in terms of IPv4 addresses per capita.  If 
>>> so, then simply removing "reciprocal," would have the same effect (of 
>>> allowing transfers to regions with more need for IPv4 addresses than 
>>> the ARIN region) and be much simpler.
>>   Region  | /8 count  | population (mm) | ipv4 per capita (+/- avg)
>>   --------+-----------+-----------------+-------------------------
>>   ARIN    |    36     |            579  | 1.043 (+355%)
>>   AFRINIC |     5     |           1216  | 0.068 (-430%)
>>   LACNIC  |     9     |            442  | 0.357 (+120%)
>>   RIPE    |    35     |            738  | 0.794 (+270%)
>>   APNIC   |    45     |           4476  | 0.168 ( -57%)
>>   --------+-----------+-----------------+----------------
>>   total   |   130     |           7451  | 0.293
>> numbers taken from
>> https://en.wikipedia.org/wiki/List_of_continents_by_population
>> Kind regards,
>> Job
>> _______________________________________________
>> You are receiving this message because you are subscribed to the ARIN 
>> Public Policy Mailing List (ARIN-PPML at arin.net).
>> Unsubscribe or manage your mailing list subscription at:
>> http://lists.arin.net/mailman/listinfo/arin-ppml
>> Please contact info at arin.net if you experience any issues.
>> _______________________________________________
>> You are receiving this message because you are subscribed to the ARIN 
>> Public Policy Mailing List (ARIN-PPML at arin.net).
>> Unsubscribe or manage your mailing list subscription at:
>> http://lists.arin.net/mailman/listinfo/arin-ppml
>> Please contact info at arin.net if you experience any issues.

More information about the ARIN-PPML mailing list