[arin-ppml] Discussion Petition of ARIN-prop-125 Efficient Utilization of IPv4 Requires Dual-Stack

[Jack Bates]

On 1/3/2011 1:43 PM, Chris Grundemann wrote:
> The same way you do today; network designs, device configurations,
> SWIP/WHOIS, other network documentation, affidavit, etc. ARIN reviews
> the documentation and then spot checks to find inconsistencies (which
> trigger a full audit if needed).
>

Device configurations? Really? Does ARIN actually ask for those? 
Honestly, a person can abuse and lie though this proposal as easily as 
if the proposal wasn't so specific. The proposal doesn't stop abuse, but 
makes engineering decisions for networks.

> Again, I am willing to include parallel IPv6 deployment.
>

NAT64 as well. I'm sure there will be other migration technologies which 
I'm not familiar with.

> How do you plan to deploy services over IPv6 without DNS?
>

Since when does an eyeball network require authoritative DNS?

>   - for every IPv4 address requested, at least one pre-existing
> interface is dual stacked, up to 80% of all interfaces and

Hi, I need a /19. How would you like me to prove the 6,554 IPv6 
interfaces? Oh, and are we proving my v6 readiness or my customer's v6 
readiness? I'm not sure their linksys is up to the task yet.

> I am much more concerned with the services that networks provide than
> how they are managed. In order for the Internet to grow, we must
> transition to IPv6. You can, of course, choose not too be a part of
> the Internet.
>

Agreed, we need v6. ARIN trying to govern network engineering with no 
possible way to handle the vast volume of data necessary to actually 
review the justification is what I disagree with.

> 1) If the IETF starts working on a superior transition method, we can
> adjust policy accordingly.
> 2) There is probably not enough life left in IPv4 for this to occur.
>

This may be correct, but I'm not willing to give ARIN more authority 
over my network and how I design it. At most, I'd let them reserve v4 
for those who are actively using v6.

> Sustainable, long-term IPv6 deployments will be native everywhere
> which that is possible. Today I see two possibilities for this;
> dual-stack and parallel network. Please point out the missing
> technology and I will happily include it in the policy.

I'm sure CGN and NAT64 and a few other methods I haven't considered will 
crop up. Heck, I won't be surprised when proxy servers are used to feed 
v4 customers v6 websites (they are today).

My objection, to simplify the above is:

1) ARIN does not have the manpower to review the large volumes of data 
to prove 80% worth of a large allocation is assigned as v6.

2) ARIN often allocates to a member similarly sized allocations (for my, 
that's normally /19-/20). The proposal says I need 6554 IPv6 addresses 
assigned, and then I can just keep asking for v4 without growing v6.

3) ARIN should not be given more control over network design.

4) Proposal will cause early IPv4 exhaustion, by causing the first 
networks to be denied IPv4 address. Exhaustion is NOT when all RIR's 
have allocated the last of their space, but when the first customers are 
denied when need is shown.



Jack