On Oct 10, 2013, at 9:34 AM, Jim Reid <[email protected]> wrote:

> On 10 Oct 2013, at 16:43, Dan York <[email protected]> wrote:
> 
>> there's nothing that DNSSEC or anything else could have done here
> 
> Perhaps that's the case for the incidents you described Dan.
> 
> However DNSSEC could help provide some form of two-stage authentication for 
> these sorts of requests. Says he hand-waving...
> 
> Some sort of token which identifies the EPP transaction could be given a name 
> and entered into the zone that's getting redelegated or whatever. That RR 
> would need to be signed. [For bonus points, the RDATA of that RR could be 
> that token encrypted with the private KSK or ZSK.] The registry checks this 
> RR before acting on the EPP request, rejects it if something is wrong and 
> raises an alarm.
> 
> This would mean an impostor would have to do more than just compromise some 
> registrar's control panel or send a fake fax. They would need to get access 
> to the zone and its keys. Which in an ideal world would be isolated from the 
> boxes a registrar uses to speak to the Internet or to the registry.

My hands can wave faster than yours:

Don't use passwords for registrant-registrar interactions, use public key 
crypto. Put a copy of the public key in a new RRtype in the signed zone. When 
the current zone owner wants to change the key (similar to a password change), 
they update that record.

--Paul Hoffman
_______________________________________________
dns-operations mailing list
[email protected]
https://lists.dns-oarc.net/mailman/listinfo/dns-operations
dns-jobs mailing list
https://lists.dns-oarc.net/mailman/listinfo/dns-jobs

Reply via email to