Re: [v6ops] I-D Action: draft-ietf-v6ops-nat64-srv-00.txt

[Martin Huněk <[e-mail]>]

Dne úterý 12. března 2019 16:19:07 CET, Philip Homburg napsal(a):
> >I would invite commentary on the substance of the draft, which has to do
> >with how one determines the prefix one should use in a CLAT, and
> >proposes a solution in an SRV record.
> 
> Let me summerize this proposal as: take the search list you get in an RA
> option (or from DHCPv6), prepend some strings and use that to look up the
> NAT64 translation prefix and other information in DNS.
> 
> At first glance it seems clever. And certainly RA/DHCPv6 agnostic.
> 
> What I'm worried about is overloading the DNS search list with network
> configuration.
> 
> Suppose an organisation or ISP has the same search list everywhere but
> some networks have NAT64 and others don't (or different networks have
> different prefixes).
> 
> You can add a dummy name to the search list, but that name will then attract
> DNS lookups.

I such case it could be solved by multiple views to the zone or higher order 
subdomain in DNSSL. Like: example.net, areaXYZ.example.net. I agree that it 
would attract DNS lookups so it might not be ideal.
> 
> Another question, how robust is the search list. Do CPEs obtain the search
> list from the upstream ISP and then announce them downstream?

Yes, this is major setback for solving DoH vs. DNS64 issue as DNSSL is not 
transitive through CPE (and it might brake things if it would be). I'll try to 
come up with some solution for that. It seems like I've been looking on that 
from just one angle.

Martin