[arin-ppml] Is Emergency action warranted for Policy Proposal 123: Reserved Pool for Critical Infrastructure?

[Owen DeLong]

On Dec 21, 2010, at 8:27 PM, Eric Brunner-Williams wrote:

> Scott,
> 
> You ask "[do I] believe proposal 123 should allow TLD operators to delay IPv6 deployment?"
> 
> If 123 is adopted, and if a similar policy is adopted in RIRs that provide allocations to "developing economies", then the existence of the policy provides a basis to argue that v6 capacity may be deferred from no later than the transition to delegation period to some point in time subsequent. Whether the deferral is for one year or three is not as important as making the policy choice that applicants will fail if, at transition to delegation, they do not have v6 capacity.
> 
> To grasp the full import, assume 1k applications, with between 60% and 80% survival to "transition to delegation", in the 2011 cohort reaching this state in 2012 and compelled to complete the transition in a fixed period. To make it worse, bear in mind that all of the 1k applicants need to have committed, when submitting their applications, in late 2010 or early 2011, their v6 capabilities.
> 
You say that like it would be a bad thing.

> If half do not use existing legacy platforms, VGRS etc., than on the order of a third of all applicant-operators are at risk of administrative failure, or of operational failure, for lack or, or lack of experience with a single point of failure, v6 capacity.
> 
It is absurd to think that people who don't have IPv6 capabilities today should be allowed to run TLDs.

> The naive v6 requirement, like several others, but this one is the only one which raises an ASO, hence an NRO, hence an ARIN policy issue, is likely to have the effect of channeling applicants to the existing legacy legacy platforms. It "raises the bar" for competition to the existing legacy legacy platforms, or the need for competencies for which existing registry operators have no contractual obligation to meet.
> 
I disagree. Most NS platforms and many non-legacy registry/registrar service bureaus are IPv6 capable now.

I think it is far more likely that this requirement will drive the others to develop their IPv6 capabilities or perish.

Again, I'm not seeing a problem here.

> As your question goes to my beliefs, I believe that existing data centers are not going to cease servicing requests for data from v4 addresses. I also believe that existing data centers are not going to be as immediately, or as critically, affected by v4 exhaustion as are access network operators.
> 
True.

> I suggest that the overwhelming majority of addresses used by domain registrants will be, during the same period that new registries are experiencing their first operational year(s) in the 2011 round of new gTLDs, v4 addresses.
> 
I think that's a poor way of stating the situation.

I suggest that the majority of addresses used by registrants will be IPv4, but, that the majority of new registrants public facing services in the 2011 round of gTLDs will also have IPv6 capabilities either immediately in 2011, or, not later than 2013.

> As to your beliefs, with respect, if "the main driver for requiring new TLD operators to be v6 capable is to ensure that new Internet users, who will have IPv6 as their primary connectivity, will be able to access the TLD nameservers over IPv6", then we, as two policy development communities, are indifferent to the interests of new Internet users in existing resources made available through a name to address mapping mechanism, and we are indifferent to the interests of existing Internet users in the organization of existing resources through a name to address mapping mechanism.
> 
That statement doesn't parse for me. It's almost as if you are claiming that requiring IPv6 somehow prohibits IPv4. Nothing could be further from the truth. Requiring IPv6 requires TLD operators to support IPv6. It makes no statement one way or the other about their support or lack of support for IPv4. I think it is unlikely any business would succeed without IPv4 in this area for several years. As such, I don't think requiring IPv4 is particularly necessary. I do think that failing to require IPv6 would show indifference. Not the other way around.

> Existing resources currently used in rural North America can be named in a new rural North American namespace without necessarily being renumbered and existing resources currently used in Africa can also be named in a new african namespace, also without necessarily being renumbered.
> 
Sure. However, new registrants will just as likely want to provide services to both IPv4 and IPv6 either immediately or in the very near future.

It is unfortunate that we cannot require the existing TLD operators and registrars to support IPv6 so easily, but, exempting new TLDs from this requirement seems absurd to me at this time.


Owen