Stability of the Internet?

Eric A. Hall ehall at ehsco.com
Fri May 18 18:03:31 UTC 2001


> Perhaps we need a tld or a group of tld's which are analogous to
> RFC1918 addresses?

This has been brought up a couple of times, but there are some pretty big
issues with it.

When somebody says "like RFC1918" you also need to include "problems with
RFC1918" in that scope. For example, private domain names allow for local
reuse of global identifiers that collide in nasty ways. What happens with
RFC1918 addresses when two orgs use the same global identifiers locally
and then need to interconnect: somebody has to renumber. The same is true
for .pri (or whatever) domain names, in that Cowboy Hats, Inc. and Cowboy
Boots, Inc. may both setup cowboy.pri domains, when they merge they have
to do a lot more work which means that any original savings (of which
there are none, if any) would have been lost.

Also like RFC1918, private domain names will leak out in unexpected ways
causing various problems. Cache poisoning was bad enough, it would become
horrific with overlapping domains.

There is some (as yet unpublished) research data that says ~20% of the
queries currently going to the root servers are for invalid TLDs (as setup
by .private internal operators). Endorsing the use of private domains will
make this much worse.

The best solution -- just like with addresses -- is to use real domains.
Advocating private domains causes more problems than it would solve.

-- 
Eric A. Hall                                        http://www.ehsco.com/
Internet Core Protocols          http://www.oreilly.com/catalog/coreprot/




More information about the NANOG mailing list