I see a lot of talk about need to engage with the TLD community and the registries.
For the record, I'm completely in agreement with Paul Wouters and Joe Abley that this is not really needed. This is why: 1. CDS will be an option for automation. There is nothing suggested saying that what we use today (i.e. out-of-band communication) should go away. On the contrary, will remain, and has to remain, because CDS cannot solve all problems (f.i. bootstrapping a secure delegation, algorithm rollover, etc). Hence, if a particular TLD doesn't like CDS, that's basically irrelevant. 2. As CDS will be an option I expect some registrants will want to use this (that's the whole point after all, that there is a need). But I also expect some registrants NOT to want to use this. Hence use-of-CDS will basically be on a registrant by registrant basis. Who has the registrant connection? The registrar, not the registry. I.e. the REGISTRAR may provide the added value service of looking for CDS and automatically convert that into an EPP transaction that goes into the already existing and fully functioning communication channel that the registrar has with the registry. The registry doesn't even have to know that CDS is in use. Hence it is irrelevant whether there's a policy that requires the registry "not to talk directly to registrants" or not. 3. CDS is a protocol feature. It is not a policy issue. There's no need for policy development. Should registries "not like CDS", that's ok (they will not know that a particular EPP update regarding a DS was triggered by a CDS). It does matter that a registrar or two is interested in this. And several registrars do appear interested, so we're fine. 4. There is a rather substantial world out there which is not part of the RRR model. Most of academia and research. Lots and lots of topology in corporate networks. All sorts of hobbyists, etc, etc. There are entire countries that don't use the RRR model. Whenever one gets to close to ICANN it seems like the entire world is shaped into a registry / registrar / registrant model. That is not true. There are other parts. Jeopardizing the possibility of automation of DS changes by going too near the registries is likely to be a big mistake. So let's try to not do that for a change. That said, there are of course lots of clever people working for registries that will provide valuable input. I'm just arguing against allowing "registry buy-in" to be allowed to become a gating factor for determining CDS usefulness. Regards, Johan _______________________________________________ DNSOP mailing list DNSOP@ietf.org https://www.ietf.org/mailman/listinfo/dnsop