Message-Id: <199312152221.RAA07205@wilma.cs.utk.edu>
From: Keith Moore <moore@cs.utk.edu>
To: Larry Masinter <masinter@parc.xerox.com>
Subject: Re: URN functionality from URLs
In-Reply-To: Your message of "Wed, 15 Dec 1993 01:56:58 PST."
<93Dec15.015659pst.2732@golden.parc.xerox.com>
Date: Wed, 15 Dec 1993 17:21:52 -0500
> a) Hostnames not only get deleted in DNS, they also get reused by
> parties that have nothing to do with the original owners of the
> hostname. That's the main reason not to use them in URNs, and hasn't
> yet been addressed.
I've been assuming that the domain name derived from a URN would point to a
URN-to-whatever lookup service, not to the host providing the file service
itself. Also, the domain name derived from a URN would be in a different
part of the tree than domain names used to look up host addresses.
The point is, the domain name derived from a URN isn't the primary name for a
host, it's just a way to find a URN->whatever server given a URN, using the
DNS.
> c) if the DNS 'works' for host names, why not replicate it rather than
> overload it? Can we use DNS protocols without using the existing DNS
> servers? Yes, every machine how has to know it's URL server instead of
> its DNS server, so we've added more configuration. But, you know, DNS
> might be replaced on a different time scale than you want URN->URL
> location service to be supplanted. Tying the two services together
> seems like senselessly making trouble.
>
> Anyway, the DNS people won't let you use DNS for anything like this,
> so you might as well stop dreaming, even though it might work.
Creating a completely separate service, even if it uses DNS protocols, makes
it much harder to get the thing going. A big advantage to using DNS is that
it's already in place.
Seems like most of the problems could be solved by rooting the URN->whatever
servers at a different part of the DNS tree than is used for normal
hostnames. (maybe in the .int domain?)
My understanding is that the big DNS problems are with scaling -- there are
too many entries under .com or .edu, the servers get bombarded and it's a big
pain to replicate that data. I don't think we would be making this any worse
if we choose the root of the URN tree carefully. (Though we would want to
make sure we don't make the same mistake in the layout of *members* of the
URN tree.)
Perhaps we could get more information from a DNS guru on the
constraints/impact of using it for such things?
Keith