To:
Mats Dufberg <dufberg@nic-se.se>
Cc:
<ietf-provreg@cafax.se>
From:
"Jordyn A. Buchanan" <jordyn@register.com>
Date:
Fri, 6 Apr 2001 16:10:52 -0400
In-Reply-To:
<Pine.BSF.4.30.0104062156300.27452-100000@spider.nic-se.se>
Sender:
owner-ietf-provreg@cafax.se
Subject:
Re: Nameserver MUST HAVE IP
At 10:02 PM +0200 4/6/01, Mats Dufberg wrote: >On Fri, 6 Apr 2001, Jordyn A. Buchanan wrote: > >> If you are the authoritative registry for a TLD and you don't have >> the IP address of a name server within your TLD, how can the NS's IP >> address possibly be found through a "normal lookup"? > >If ns1.foo.com and ns2.foo.com are the nameservers of foo.com then the IP >addresses is a must. > >If then ns3.foo.com and ns1.foo.se are the namesevers of bar.com, neither >of the IP addresses are needed (the IP address of ns1.foo.se is forbidden >in com zone). The IP address of ns3.foo.com is to be found in the foo.com >zone through normal lookup. And if foo.com's name servers are in bar.com at the same time? I realize that this is an extreme example, but this is probably not something you want to allow. >If you don't think it works, I'll have to find some exampel from the SE >zone. :-) I think it works sometimes, but that doesn't make it a good idea. All of this comes from treating name servers as attributes of domains rather than objects. If you think of the name server as an object in the registry as opposed to an attribute of SLD.se, then the requirement of IP addresses for name server objects within the TLD makes more sense. I realize that some registries today don't take this approach, but the fact that it also happens to prevent loops like the one I describe above is simply another argument in favor of the name-server-as-object approach. Jordyn -- Jordyn A. Buchanan jordyn@register.com Futurist +1.212.798.9262 Register.com http://www.register.com/