Re: new port for DNS

Dirk.vanGulik (Dirk.vanGulik@jrc.it)
Mon, 19 Jun 1995 10:29:18 +0200


Date: Mon, 19 Jun 1995 10:29:18 +0200
From: "Dirk.vanGulik" <Dirk.vanGulik@jrc.it>
Message-Id: <9506190829.AA01487@ jrc.it>
To: dupuy@smarts.com, uri@bunyip.com
Subject: Re: new port for DNS

Correct me if I am worg, but I strongly feel that this DNS
discussion could perhaps be sidestepped:

*WHEN* talking about a resolving mechanism, we want to resolve
a URN. In the DNS naming scheme we have a DNS name which specifies
the host which can resolve the URN for us. 

So we have *TWO* tasks; 

	A   resolving the host-name (through DNS)
	B   asking that host to resolve the URN.

Assume the <urn:dns:food.bar.eat.it:my.very.boring.doc.txt>

The existing dns system helps you to the IP address for food.bar.eat.it,
possibly by going al the way to a name server in the eat.it domain
which tells you the IP address.

With this IP address you start talking to the host at a certain well
known port which will resolve the 'my.very.boring.doc.txt' for you.

In short, with the DNS scheme (and any others IMHO) there is *no* need
for duplication the DNS resolving mechanism in the URN resolving 
mechanism. We'll just be talking to *two* ports; one to get the hostname
to IP address which gets us the resolving host, and the other one
to resolve the actual URN.

Dirk-Willem van Gulik

http://me-www.jrc.it/~dirkx               http://ceo-www.jrc.it
DWvGulik@Dialis.xs4all.nl                  Dirk.vanGulik@jrc.it
+39 332 78 1322                                 +39 332 78 9549
                                            fax +39 332 78 9185

ISEI/ESBA;                     The Center For Earth Observation
Joint Research Centre of the European Communities, Ispra, Italy