[rpd] Off-topic (was Re: [PDWG-Appeal] REPORT ON Appeal against the non-consensus determination on proposal AFPUB-2019-GEN-006-DRAFT02 (RPKI R)

[David Conrad]

Owen,

On Mar 12, 2021, at 3:34 PM, Owen DeLong via RPD <rpd at afrinic.net> wrote:

> Yes, to some extent, it is true in that there was an internet governance system in place prior to the creation of LACNIC and AFRINIC. However, RIPE-NCC and APNIC almost entirely predate anything that could reliably be called an internet governance system (basically being part of the early stages of a transition off of IANA being a notebook maintained by Jon Postel). ARIN came next and IMHO was built somewhat contemporaneously with the idea of any sort of formal system of internet governance.


Depending on how you define it, “Internet governance” existed long before any of the RIRs. RIPE-NCC and APNIC were both established under the concepts defined in RFC 1366/1466. For an academic, exhaustive, extensively footnoted (480!), treatment on the issue see https://www.icann.org/en/system/files/files/creation-administration-unique-identifiers-1967-2017-18nov20-en.pdf.


> However, in the finite number space, there is no way for this to happen in a mostly non-overlapping or non-conflicting

> manner and unlike DNS zones where you can choose your TLD operator on the basis of who cares about your zone,

> when it comes to uniqueness of IP addresses, having different ISPs regarding different registries as authoritative will

> be an adventure nobody will enjoy.


To be clear, it isn’t a question of the finiteness of the number space (the name space is also finite, although much larger), it’s a question about how the uniqueness is implemented. In the namespace, it is done via the resolvers, typically root hints/KSK. In the numbering space, it is done via announcement made by network operators. The key difference is that in the namespace, name uniqueness imposed by a resolver only impacts those who use that resolver. In the case of numbering, at least historically, announcements are propagated beyond the resolver-equivalent (i.e., edge routers), so the numbering equivalent of “name collisions” is detectable much earlier and can have more impact. With the deployment of RPKI, the model changes somewhat and the fact that all 5 RIRs now claim 0/0 increases the risk of “alternative roots” with potential conflicts in the (wildly unlikely) event that cooperation of the RIRs break down.

Of course, this is unrelated to the topic at hand.

Regards,
-drc

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.afrinic.net/pipermail/rpd/attachments/20210408/550deec7/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: Message signed with OpenPGP
URL: <https://lists.afrinic.net/pipermail/rpd/attachments/20210408/550deec7/attachment-0001.sig>