To:
ietf-provreg@cafax.se
From:
Kent Crispin <kent@songbird.com>
Date:
Tue, 6 Feb 2001 09:19:59 -0800
In-Reply-To:
<IPEMICCPDPPICMIONJIOEEACCCAA.briansp@walid.com>; from Brian W. Spolarich on Tue, Feb 06, 2001 at 10:02:47AM -0500
Mail-Followup-To:
ietf-provreg@cafax.se
Sender:
owner-ietf-provreg@cafax.se
Subject:
Re: Nameserver as object/entity or not ?
On Tue, Feb 06, 2001 at 10:02:47AM -0500, Brian W. Spolarich wrote: > | I'm working in a Registrar now, and it happens that customer ask us > | to change 50 or 100 domains at the same time because they change > | nameservers or they change IP of one nameserver. > > This is a good point, and my mindset tends to not be focused on > service-provider concerns as much as registrant concerns (somewhat ironic > given I used to work for a rather large ISP). ISPs who renumbered would > have an impossible time if they had to coordinate a massive batch of updates > to the registration entries for all of their customer domains for which they > were providing nameservice. > > I would say that this concern alone would warrant a requirement that the > nameserver entities be managed separately. Though apparently there are registries that do not provide this service, and things seem to work... In any case, what you describe is a requirement that there be a way to manipulate all records that include a particular nameserver. It is not a requirement that nameserver objects should exist -- that is an implementation detail. If you have nameserver objects, then you have in effect created potentially conflicting ownership/authentication domains, and this seems to me a very fundamental problem. (That is, a domain can be "owned" by A, while the nameservers for the domain may be owned by "B".) If only domains can be owned objects, then there are no possible conflicts (at least in the registry DB). -- Kent Crispin "Be good, and you will be kent@songbird.com lonesome." -- Mark Twain