Message-Id: <9307241637.AA03183@tipper>
To: Rob Raisch <raisch@ora.com>
Subject: Re: URNs in the DNS
In-Reply-To: Your message of "Fri, 16 Jul 93 13:12:10 EDT."
<Pine.3.03.9307161308.A1469-b100000@amber.ora.com>
Date: Sat, 24 Jul 93 12:37:48 -0400
From: Simon E Spero <ses@tipper.oit.unc.edu>
[alan: please re-attach me to the uri mailing list. I promise to be good :-)]
Hi Y'all.
I just got back from Europe, and I missed some of the earlier discussion,
so some of this may be going over old ground.
I printed out a copy of the DNS spec before I left Amsterdam, and had a long
look through it to see what would have to be changed to create a system
capable of handling URN mappings. The conclusion that I came to is that
any reasonable solution would require extensions within the protocol,
together with quite different implementation techniques for servers.
Apart from the problem of tying URN resolution to the place where the URN was
conceptually issued (making caching of documents very difficult), all
implementations of DNS that I know of use memory resident databases.
Whilst memory residency made sense when running on a Vax serving small
amounts of data, this approach cannot reasonably be scaled to large datasets.
With advances in hardware and operating systems, it should now be possible
to change implementation strategies to be more disk-with-cache based whilst
still offering the required performance.
I'll try and write up some of my suggestions for DNS++ (tm) later
Simon