Multiple Roots are "a good thing" - Karl Auerbach

Greg A. Woods woods at weird.com
Sun Mar 18 20:47:04 UTC 2001


[ On Sunday, March 18, 2001 at 14:23:26 (-0500), Miles Fidelman wrote: ]
> Subject: Re: Multiple Roots are "a good thing" - Karl Auerbach
>
> I would suggest that telephone books/directories are not an appropriate
> analogy. Rather, DNS is a lot closer to the internal plumbing of the net -
> more akin to Signalling System #7. I'd guess that for 95% or more of phone
> calls, the caller already knows the numeric phone number in question -
> while for the Internet, very few people give their email addresses as
> mfidelman at 207.226.172.79 or http://207.226.172.79. Telephone directories
> are optional in most cases, DNS is not.

You are absolutely correct.  :-)

Telephone directories are most definitely *not* like the DNS.  A domain
name is more like a telephone number itself, and as you say the IP
numbers are more like the underlying circuit routing glue in something
like SS#7.  We really do not have a "telephone directory" for the
Internet (unless you count WHOIS/RWHOIS).  A directory is something that
can be searched with approximate matches.  Because the DNS is
"D"istributed, it is literally impossible search it that way (and if
there were multiple roots then all users would really be up the creek
without the proverbial paddle!).

> Yes, the Internet can function on
> numeric IP addresses alone

The Internet could sort of run on IP address numbers alone (but it
almost never has -- there was hosts.txt before DNS).  However since IP
numbers can change (at a much greater frequency than telephone numbers
ever change) without the "content" changing, the indirection of DNS
names to IP numbers is a critical part of the longer-term consistency of
the net.

In the phone system analogy it would be like having the phone company
come along and randomise your entire number every month or so (not just
your "network" (aka local exchange) number).  Since most people don't
actually move locations that often such a regular but random renumbering
that was not in the direct control of the user would cause general havoc
with telephone users.

In the real-world phone systems it might not be so easy to re-number and
re-route exchanges in the underlying signaling systems as it is to
renumber IP networks, but then again most analogies only go so far....

> - but unlike the phone network, people don't
> give out email addresses or URLs containing their numeric host addresses. 

Well, some spammers do, but that's their fault!  :-)

> In other words, we need an authorized international body with the clout to
> oversee the whole mess. But then, isn't that what ICANN is supposed to be?
> (Or would you rather have the ITU oversee the Internet?)

some days I'd rather have the UN do it...   :-)

-- 
							Greg A. Woods

+1 416 218-0098      VE3TCP      <gwoods at acm.org>      <robohack!woods>
Planix, Inc. <woods at planix.com>; Secrets of the Weird <woods at weird.com>




More information about the NANOG mailing list