W3C home > Mailing lists > Public > ietf-discuss@w3.org > November 2001

Re: Request for comments

From: Claudio Allocchio <Claudio.Allocchio@garr.it>
Date: Fri, 16 Nov 2001 10:22:13 +0100
To: Jasdip Singh <jasdips@erols.com>
cc: Keith Moore <moore@cs.utk.edu>, discuss@apps.ietf.org
Message-ID: <Pine.VMS.3.91-B.1011116100704.24344A-100000@SYNX02.elettra.trieste.it>

> > the question is not whether this will work, but whether it's better to
> > put stuff in DNS or to vector it to a separate lookup service.

... this is a very old discussion, and in general the opinion was DNS 
works well, but it is also the fundamental glue between addresses and 
services, names etc, thus overloading it with new stuff IS WRONG.

More over I believe that DNS is OK for machine driven quick lookups, i.e.
domain name to server address or server name. 

> protocol? The dreadful question IMHO seems to be where do you stop touching
> DNS? Also, where to start and stop using directory services like LDAP?

One of the main resons for using DNS is often said to be "it works, it is 
everywhere". But:

 - this is no nore exectly true: it "seems to be everywhere", but virtual
   hosting often makes the DNS server not under control of the domain name

 - it works, but it is defintly not suited for an "information lookup 
   service". IMHO the major disaster that DNS has ever made was the attempt
   to identify it with a trademark and distighushed names database. :-)

 - currently, and finally, due to a number of application seriously needing
   real directory services in order to work (GRID computing, mobile users
   authorisations - AAA, on-line phone books, X.509 Government issued
   certificates to citizens,...) the LDAP / Directory infrastructure is
   being deployed (often non hyerarchically, but the directory bridge concept
   seems to overcome the problem). This infrastructure is not at all in the
   same places and same hands than DNS servers, but it seems much more
   suitable to the needs expressed in your document.

It is just a personal view, of course, but I woul prefer efforts being 
aimed to coordinate LDAP servers and define the relevant OIDs.


Received on Friday, 16 November 2001 04:23:11 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:38:01 UTC