Re: DNS under o=Internet

yeongw@psi.com Tue, 11 February 1992 14:30 UTC

Received: from nri.reston.va.us by NRI.NRI.Reston.VA.US id aa09057; 11 Feb 92 9:30 EST
Received: from bells.cs.ucl.ac.uk by NRI.NRI.Reston.VA.US id aa09053; 11 Feb 92 9:30 EST
Received: from psi.com by bells.cs.ucl.ac.uk with Internet SMTP id <g.13940-0@bells.cs.ucl.ac.uk>; Tue, 11 Feb 1992 14:05:49 +0000
Received: from localhost by psi.com (5.61/2.1-PSI/PSINet) id AA02010; Tue, 11 Feb 92 09:05:23 -0500
Message-Id: <9202111405.AA02010@psi.com>
To: George Michaelson <G.Michaelson@cc.uq.oz.au>
Subject: Re: DNS under o=Internet
Cc: osi-ds@cs.ucl.ac.uk
Reply-To: yeongw@psi.com
In-Reply-To: Your message of Tue, 11 Feb 92 11:29:31 +1100. <28255.697768171@brolga.cc.uq.oz.au>
Date: Tue, 11 Feb 1992 09:05:22 -0500
From: yeongw@psi.com

>   Wengyik, how do we take the existing non-US located DNS information
>   and get a complete and consistent mapping into the DIT if we have to
>   take these to each individual DMD for the given ISO countrycode, and
>   get it authorized under the civil space?
> 
>   You just relegated all non-US located DNS people into 2nd class.

I don't understand. This has nothing to with the U.S. What I'm suggesting
is simply that people continue to be listed in the part of the DIT
containing the civil namespace, and the DNS tree be located in another
part of the tree. 

I'm also suggesting, as a separate issue, that pointers be constructed
between the two parts of the DIT to make searching more efficient
than it currently is.

>   Apart from that, the labour is going to be massive.

I don't understand this either. What labor? To construct the associatedName
pointers? That can be done automatically by doing searches for
associatedDomains in the White Pages part of the tree (which 'fred',
among other UAs, already does -- I'm just proposing to do it once, when
constructing the tree and then save the results, instead of doing it
every time a search is performed).

>   Apart from that #2, civil servers and Internet servers being distinct,
>   this means a LOT more chain/referral for simple DNS lookups. 

No. Simple DNS lookups don't ever have to look up anything outside the
DNS part of the tree. The only time the associatedName pointers need
to be followed is for 'hybrid' lookups involving both DNS and
White Pages information.

The 'hybrid' lookups are added functionality that you currently
can't do with either the DNS (White Pages information not present)
or the current White Pages-centric DIT (no DNS information).

>   It ain't gonna be fast. It ain't gonna do reverse-IP lookups well.
> 

White-Pages only or DNS-only lookups aren't going to be any slower
than they are currently. Now admittedly that's incredibly slow to
begin with, but that's another problem.

The 'hybrid' lookups are going to be slower. But it's also added
functionality. Which would you rather have, slow functionality or
no functionality?


Wengyik