[arin-ppml] Proposal to ban Leasing of IP Addresses in the ARIN region

[Owen DeLong]

> On Sep 22, 2021, at 11:52 , William Herrin <bill at herrin.us> wrote:
> 
> On Wed, Sep 22, 2021 at 11:32 AM Mike Burns <mike at iptrading.com> wrote:
>> Every LIR is a mini-ARIN by nature, isn't it?
> 
> No. And yes. And no.
> 
> I personally dislike the term "Local Internet Registry" precisely
> because it creates this ambiguity. I'd be happier if we just stuck
> with "ISP."

Problem is ISP doesn’t really work when you include the myriad other
kinds of address providers that have to exist in the modern world, such
as cloud, hosting, colo, etc.

> To my point of view, a network service includes IP addresses. The ISP
> isn't really acting as a mini-ARIN, they're providing a network
> service.

If that were true, this wouldn’t be an issue, but with various ISPs now
charging a separate per-address fee, why shouldn’t they be subject
to competition for that aspect of their business separate from their
connectivity business? Why is it ARIN’s place to shield them from 
competition?

If you want to talk about an anti-trust issue, wouldn’t this be the classic
case?

> When they provide so many IP addresses with the service that it
> becomes a number policy concern then you could say they're acting as a
> mini-ARIN. Which I think is a problem. There's a long-standing
> practice of ISPs assigning /24s and more to end-users which then find
> their way into the BGP table disaggregated from the ISP's allocation.
> That troubles me almost as much as the folks who want to be
> straight-up mini ARINs without providing network services.

We all get it that you have long thought ARIN should be the rout
aggregation police. They aren’t, and that’s not going to happen.

Can we try focusing on reality as it currently stands?

> I actually did a 10-minute presentation at an ARIN meeting in Atlanta
> years and years ago where I talked about forward-looking developments
> in routing technology and the impairment that ISP disaggregates could
> impose.

And then memory got bigger, faster, cheaper and in a few more Moore’s
law iterations, will probably permanently surpass bop table growth.

Owen