[arin-ppml] the bad 240/4 idea, was Solving the squatting problem

[Owen DeLong]

> On May 17, 2019, at 12:53 , Michel Py <michel at arneill-py.sacramento.ca.us> wrote:
> 
>> Owen DeLong wrote:
>> IPv6 unicast addresses were allocated to ARIN for allocation and assignment. 
>> The addresses were in ARIN inventory and there was an erroneous view that
>> IETF had authority to dictate registration policy for those addresses.
> 
> That was definitely not the view of the IETF. I can remember some private discussions that were not exactly politically correct. You did good on that one, we were so stuck into that multi6 mess.

Regardless of the opinion of the IETF (or more accurately certain participants in
the IETF, since the IETF never made a coordinated statement of any sort of
WG conclusion that I was off base on that), those were the facts on the ground.

IETF had allocated 2000::/3 to IANA to dole out as Unicast address space.

IETF had provided recommendations about how it was to be managed, but whether
IETF (or more accurately certain IETF participants) recognized the limits of their ability
to enforce those recommendations, the RIRs were, in fact, free to make policy regarding
the administration of those address spaces (individually for space allocated to them by
IANA and collectively (via the ASO AC) for how IANA would manage allocations to RIRs).

That’s been the structure of address policy for v4, v6, and ASNs since IANA first started
delegating address management to RIRs.

>> David Farmer wrote :
>> If you are a believer in IPv4-Only
> 
> I'm not a believer of anything, I just do whatever I can or must. I'm leaning that way tough, because I'm getting tired of people trying to shove IPv6 down my throat when I can't afford it.

You keep saying this as if there was some huge expense to IPv6. I admit I don’t know your specific network, but it’s pretty hard to fathom that at this point.

> I have DECNET on my production network. I have HPUX, VMS, Netware, OS/2 Warp, and any flavor of Linux and Windows you can name. I have T1s. I don’t get to pick what's on the network, I just have to make it work.
> 
> I know what IPv6 is. I was on the 6bone. I hate to break it to you, but at $job[0], it's not even on the agenda.

Then there’s probably a limit to the amount of time that $job[0] will remain able to communicate with the full internet.

You quoted 25% and 3% numbers earlier, I don’t know what your source is, but from Google, Facebook, Netflix, and Akamai, it looks like something close to 50% of all internet traffic in North America is IPv6. More than 50% of all mobile traffic in that same area. World wide the numbers drop some, but they are continuing to rise.

> We are heading straight towards the balkanization of the Internet.

That’s been true to some extent since the early days of IPv6 development. There was always going to come some time when the majority of the internet was going to decide that supporting IPv4 wasn’t worth the cost or hassle any more and at that point, you’d have IPv4 islands in an IPv6 ocean. That fits the definition of balkanization[1] for some.

Owen

[1] Apologies to Jan Zorz for using your chosen term, which is horribly politically incorrect in his opinion, and he should know.