Re: Last Call: <draft-iab-2870bis-01.txt> (DNS Root Name Service Protocol and Deployment Requirements) to Best Current Practice

[David Conrad <[e-mail]>]

Hi,

I've been biting my tongue a bit on this discussion, but have lost the battle.

On May 28, 2014, at 12:18 PM, Michael Richardson <[e-mail]> wrote:
>>> I would like to see each and every root server support both IPv4 and IPv6.
>> So would I. But is that a *requirement*, particularly given that the
>> root service seems to run just fine today without it?
> 
> I would like every A-M.root-servers.net have an A and AAAA record.
[...]
>   The root name service:
>      ...
>      MUST support IPv4[RFC0791] and IPv6[RFC2460] transport of DNS
>      queries and responses.

And if they don't?

Just for clarity, the root server operators are under no obligation to do anything. The whole "MUST" bit is actually sort of misplaced since it isn't like people are going to wave a RFC (BCP or otherwise) at the root operators and change will magically happen. Root server operators will do what they want according to their own requirements/business drivers. In an ideal world, what the community wants and what the root server operators' requirements/business drivers are correspond, but people shouldn't be under any illusion that an RFC will make this happen.

> because it focuses on the root name *service* from an external functional
> point of view, and not from a design point of view.

This is probably backwards. A design document is probably a better goal since it is aspirational. A document that attempts to define root service isn't particularly useful if there is no practical way to enforce that definition.

Regards,
-drc