On 30 May 2014, at 20:24, David Conrad <drc@xxxxxxxxxxxxxxx> wrote: > On May 30, 2014, at 2:49 AM, Joe Abley <jabley@xxxxxxxxxxx> wrote: >> The current text talks about the service, not individual servers. >> >> The service (as provided by thirteen root servers) is available over both IPv6 and IPv4 today, and hence complies with the proposed BCP. > > So, to reduce this to the absurd, if a single root server operator provided IPv6 DNS service via a dust-covered 8088 PC connected over a 300 bps nailed up SLIP line, the proposed BCP would be met? > > BCP177 would appear to be a significantly better answer here. My understanding here is that this is an attempt to codify a set of protocol requirements for the service, deliberately leaving operational considerations like deployment using anycast, capacity and capabilities of individual root servers, etc to be described in a kind of self-declaration by the root server operators themselves. This was originally proposed to meet the kind of challenge you outlined in a previous e-mail relating to the odd, non-contractual relationship between individual root server operators and, well, everybody else. I think this sub-thread is mainly concerned about whether that is a sensible approach, and the particular issue of whether IPv6 transport for individual servers belongs in this document is just an example of the wider issue. Perhaps pulling up a thousand feet and considering whether the proposed split of 1. protocol requirements for the service -- RFC series 2. description of the operational aspects of the service -- RSSAC document with the idea that both would functionally replace 2870 is what needs to happen? (I appreciate, as others have mentioned, that the RSSAC document has not yet been published in draft or final copy.) Joe
Attachment:
signature.asc
Description: Message signed with OpenPGP using GPGMail