[arin-ppml] Change of Use and ARIN (was: Re: AFRINIC And The Stability Of The Internet Number Registry System)

Owen DeLong owen at delong.com
Wed Sep 15 12:44:09 EDT 2021



> On Sep 14, 2021, at 22:50 , scott at solarnetone.org wrote:
> 
> 
> 
> On Tue, 14 Sep 2021, Owen DeLong wrote:
> 
>> 
>> 
>>> On Sep 14, 2021, at 22:42 , scott at solarnetone.org wrote:
>>> 
>>>> Nobody I know has found a way to do lossless packing of 128 bits into a 32 bit field yet. Until you can achieve that, compatibility is rather limited.
>>>> 
>>>> Please present your solution here.
>>> 
>>> Encode it in four sequential packets, 32 bits per, and add logic to parse those malformed addresses in the routing daemons.
>> 
>> Either I’m missing something, or that’s not going to be functional when those 4 packets reach the IPv4-Only end host and it has to reply.
> 
> Maybe, but that is not the challenge you presented:)

Fair enough… In context, the challenge I presented was about getting an IPv4-only host with no changes to software to be able to engage
in bidirectional communication with remote hosts that live in a 128 bit address space. Yes, you are correct the the way I abbreviated my
expression of that particular challenge was not complete in itself without the additional context.

> Seriously, some manner of stateful 6/4 nat or header mangling is going to be required upstream of the legacy device to translate.

Yeah, but because of the way IPv4 has been implemented (protocols that embed addresses, expectations of dealing with rendezvous
hosts, NAT traversal assumptions, etc.), it turns out that evenstateful 6/4 NAT is unnecessarily hard and unreliable at best.

Owen



More information about the ARIN-PPML mailing list