Re: [DNSOP] [[e-mail]: I-D Action: draft-grothoff-iesg-special-use-p2p-names-00.txt]

[Joe Abley <[e-mail]>]

On Monday 2 December 2013 at 12:44, Ted Lemon wrote:

> On Dec 2, 2013, at 11:29 AM, Joe Abley <[e-mail] (mailto:[e-mail])> wrote:
> > > RFC 6761 has IETF consensus, and does not propose adding new namespaces under .arpa, but rather at the top level. Here's what RFC3172 says on the topic of .arpa:
> > > 
> > > This domain is termed an "infrastructure domain", as its role is to
> > > support the operating infrastructure of the Internet. In particular,
> > > the "arpa" domain is not to be used in the same manner (e.g., for
> > > naming hosts) as other generic Top Level Domains are commonly used.
> > 
> > Perhaps 3172 needs to be updated to reflect current IAB practice, then. It's not hard to find examples of names under ARPA which contradict the text you quoted (e.g. see RFC 5855).
> 
> This seems like a non-sequitur, since RFC 5855 refers to a function that RFC 3172 specifically mentions.

I was perhaps being a little opaque; RFC 5855 specifies names under .ARPA for hosts (A.IN-ADDR-SERVERS.ARPA, etc.) This seems to be in contradiction to the text you quoted.
> > > Aside from the purely practical matter that having special domains live under .arpa would be more complicated to implement,
> > 
> > This feels like an over-general pronouncement that can't hope to be accurate in all cases. Why is it more complicated in the general case? More complicated than what?
> 
> If special-use names are at the top, then you can just look at the terminal label to see that you need to use a different protocol to resolve names under the special-use name.

Oh, I see what you mean now, thanks. I don't necessarily agree that it's significantly easier to check the final label (e.g. ".local") than the final several labels (e.g. ".local.apple.com") but I understand your point.
> > I think you're sharing personal opinion rather than citing fact ("make sense"). The .local convention happened to be adopted by Apple for use in a DNS-like protocol, and was documented (and the IN-class top-level label reserved) later. They could equally well have adopted .local.arpa or .local.apple.com (http://local.apple.com).
> 
> So, in what sense is MDNS "internet infrastrucure?"

It's a protocol that is used on the Internet. It's a protocol that is used to locate and obtain addresses for hosts connected to the Internet. I appreciate that it has a restricted scope, but in practical terms so does everything.
> > > The other proposed special uses are similar. Putting them under .arpa might be _expedient_, because it avoids the whole change control question, but that's pretty much the only way I can think of that it makes sense.
> > 
> > It doesn't avoid the whole change control question -- it just reduces the change control to one with a single, uncontentious decision point (the IAB). By contrast, the business of identifying reserved strings (and enforcing their non-delegation for other purposes) in the root zone is fraught with administrative ambiguity.
> 
> It's not at all clear to me that this is an issue, and if it is, we should figure that out.

I don't disagree. It's not clear to me either; I'm just conscious of the fact that perspectives can vary depending on whether you're used to looking at things from an IETF perspective or an ICANN perspective. I have been somewhat familiar with both, and I'm confused. :-)
> I would like to think that ICANN understands that there is a change control conflict here, and is willing to work with us to make sure that we don't step on each other's toes. Clearly it's incumbent upon the IETF not to be frivolous in the use of RFC 6761. But I have not heard from anyone that the IETF should not ever use RFC 6761, and that seems to be the position you are advocating at the moment.

I can see how I might have given that impression, but it wasn't intentional. 

I have some personal dissatisfaction with the approach taken in 6761 (I think there was an opportunity to unify a range of technical policy and collapse a number of registries that give the appearance of doing the same thing in different places, and I think its a shame that the opportunity was not taken) but I have no objection in principle to the idea that some right-most labels could/should be reserved for use by the IETF.

The reason my knee is jerking is that I think in general it's a mistake to assume that a unique top-level label is the only practical recourse for protocol designers looking for stable, dedicated namespaces. There are lots of other options, several of which mentioned in this thread, and one size does not fit all.


Joe