[arin-ppml] TIPTOP

[Daryll Swer]

Speaking of endpoints, this might not be a popular opinion, but perhaps
DHCPv6 should be forgone for space endpoint networking. We already have
enough complexity with DHCPv6 HA/state transfer/failover on Earth in
everyday residential ISP networking for M:N BNG design.

Since IPv6 is native and NAT-less, we could just use *routed* IPv6 to the
host. In a DC environment, that would mean a BGP-to-the-host design that
I'm sure many of you are well familiar with.

For *wireless* networking in space, perhaps an IGP *variation* *inspired*
by ES-IS (not IS-IS) and the Babel routing protocol would work, assuming it's
possible to design for extremely high-latency networking and limited
bandwidth.

*--*
Best Regards
Daryll Swer
Website: daryllswer.com
<https://l.shortlink.es/l/a274a18c4794eb252540be4a22756944d387de0b?u=2153471>


On Wed, 20 May 2026 at 22:41, Daryll Swer <contact at daryllswer.com> wrote:

> A multi-stakeholder working group is likely a good idea.
>
> Agreed. I don't think we should limit this discussion only to
> networking-centric people (TCP/IP/MPLS/SRv6 folks).
>
> Where would this new working group ideally be 'hosted', though? It doesn't
> seem ideal to map such a unique collaborative group to ARIN or the IETF.
> Because the group is more than just "IPv6", it would involve people with
> deep-space expertise across different scientific domains and knowledge well
> beyond TCP/IP/UDP/MPLS/SRv6/EVPN.
>
> I also believe it is worthwhile to not include IPv4 in any of these plans
>> as we’re at exhaustion and the v4 space is not nearly vast enough to
>> accommodate these needs.  I believe most of these space based
>> implementations will require unique networking to the point where IPv6-only
>> will be the least technical challenge here, especially with this and so
>> much of it being so future-focused.
>
> Couldn't have said it better myself. If we implement IPv6-native for
> space, we should make it an easy-to-use technology. We should agree on the
> addressing/subnet model for space from the start and probably also consider
> the possibility that *endpoints *like laptops, phones, etc. outside Earth
> may use an *RFC9663*-like implementation so that space-related
> organisations won't hesitate to adopt IPv6.
>
> *--*
> Best Regards
> Daryll Swer
> Website: daryllswer.com
> <https://l.shortlink.es/l/c34e3084974961b681d7c54e1ce4efbd7e2353f1?u=2153471>
>
>
> On Wed, 20 May 2026 at 21:55, Preston Ursini via ARIN-PPML <
> arin-ppml at arin.net> wrote:
>
>> I agree with many of the points raised and I believe any address
>> allocation that revolves (orbits?) around celestial body aggregation will
>> be difficult and would merit input from a body such as the IAU, Planetary
>> Society, government space agencies, commercial operators, and many others
>> with cross domain knowledge. A multi-stakeholder working group is likely a
>> good idea.
>>
>> In our own solar system we go from planet to moon and then we have dwarf
>> planets, asteroids, and many other classifications of celestial objects.
>>
>> One thing we run into is celestial objects are constantly being
>> discovered that could need address space.
>>
>> A simple hierarchy from Planet -> Moon won’t work, especially with
>> thousands of trans neptunian objects alone, and millions of other
>> categorized objects as well.
>>
>> This is only our own solar system, interstellar probes could also
>> potentially need to be addressable.
>>
>> I believe a prefix for each planet may be a good idea, but which planets,
>> of what size, and what classification of planet (dwarf / gas giant with
>> many moons / etc) is just barely scratching the surface of what is needed
>> here.
>>
>> If we really were thinking of the far future, I believe that the solar
>> system and objects only exiting the solar system such as the voyager probes
>> should be separately allocated from interstellar objects that may need to
>> one be addressed.
>>
>> On a very high level I’d propose:
>>  - Earth orbit and Near Earth Objects (LEO / GEO / Lagrangian Points /
>> etc)
>>  - Solar system bodies and spacecraft (Moon / Mars / Jupiter / Asteroids)
>>  - Interstellar probes or future non-solar system networks (Anything
>> destined to leave the solar system)
>>
>> Each of these class of objects have different needs with latency /
>> connectivity / etc, and I believe is a good start based on density of use
>> and future planning.  Moon/Mars Colony can then be addressed different than
>> an LEO satellite, and Voyager type spacecraft would be in a different range
>> from everyone else.
>>
>> There will obviously need to be major discussions as to what ranges each
>> body / orbit / etc gets.  I also believe it is worthwhile to not include
>> IPv4 in any of these plans as we’re at exhaustion and the v4 space is not
>> nearly vast enough to accommodate these needs.  I believe most of these
>> space based implementations will require unique networking to the point
>> where IPv6-only will be the least technical challenge here, especially with
>> this and so much of it being so future-focused.
>>
>>
>> Preston Louis Ursini
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.arin.net/pipermail/arin-ppml/attachments/20260520/66a67964/attachment-0001.htm>