Re: [weirds] I-D Action: draft-blanchet-weirds-bootstrap-ianaregistries-00.txt

[Byron Ellacott <[e-mail]>]

On 22/06/13 3:20 AM, "John Levine" <[e-mail]> wrote:

>>How would a client bootstrap a query for 29.12.202.in-addr.arpa ?
>
>It would look at arpa.domains.rdap.arpa to get the name of the RDAP
>server for .arpa, if there were one.  I expect that's not the answer
>to the question you meant to ask.

Presumably, the server for .arpa is then going to redirect the client to
the appropriate RDAP server for the specific resource associated with the
reverse domain being queried.  The implicit questions are: how the server
for .arpa is going to do that; and why wouldn't a client skip all the
bothersome XML parsing stuff and just ask arpa.domains.rdap.arpa?

>If you're asking how to bootstrap a query for 202.12.29.0/24, or
>2001:0400::/22, my proposal is that the numbers bootstraps are built
>into the clients, with informal advice to update them every year or
>two, sort of like the root zone hints in a DNS cache.  Given how
>slowly the IANA numbers tables change, I think that's more workable
>than something that tries to boostrap the rather irregular numbers
>assignments on the fly.

I don't mind this approach; is there suitable RFC wording that would allow
clients to have confidence that asking one of the numbers bootstrap
servers will get you to the right place for the answer you seek?

(ie, the RIRs will redirect to each other, but should that information go
into an RFC, and if so, how?)

>* Numbers assignments move among RIRs, so the RIRs have a history of
>cooperation and awareness of each other's assignments.

As you pointed out a few messages ago, numbers query clients are going to
rely on the RIRs cooperating for redirection no matter what bootstrap
method is used.

  Byron