Re: InterNIC Reliability Issues <long>

[Richard Stiennon <richards@netrex.com> (by way of Ivan Pope)]

At 11:55 PM 6/24/96 -0700, Shadow wrote:
>
>Does the InterNIC have multiple links to the 'net?
>
>This evening our site lost connectivity (It was for at least an hour to
>my knowledge) to the InterNIC's servers (web server, the root server
>housed there, etc.)...  Since our site is fed by three different
>providers (CWI, Agis, and Sprint), I find it hard to believe that we
>couldn't reach the NIC's machines.


The following was posted to com-priv@psi.com yesterday.  Karl has obviously
given all of the issues discussed here a lot of thought:



Topic: DNS Top Level Domains -- multiple registries and zones still being
       obstructed.    If you use the Internet, or do commerce over it, READ
       THIS!  This is VERY IMPORTANT!

Generically:
	There is a strong perception in the community that we need more
	registries than those which handle ".COM" (ie: ".BIZ", ".CAR", etc).
	Some of the self-appointed "governing bodies" are trying to see to
	it that either excessive cost or taxes are imposed on the users of
	the Internet to provision these resources, or prevent them from
	being deployed at all -- and at any cost.

	What follows is my commentary on the lastest draft (to get copies,
	ftp to "ds.internic.net" and grab anything with "itld" in the title;
	this can be done with "mget *itld*")

Folks, I've read every IETF draft which contains the string "itld" in the
subject line (mine, the original, and Postel's latest) and frankly, I still
don't see what the fuss is about.

Jon's draft seems to suffer from the same fatal flaws it had when it was
first proposed, to wit:

1)	It is fundamentally a tax authority.  Still.  1% of operating
	revenues, a $10,000 initiation fee, and a $1000 *NONREFUNDABLE*
	application fee?  All with the knowledge that only 150 new TLDs will
	be allocated in the first year, and no more than 2/3rds from any one
	country (presumably the US), with no more than 50 new registries?

	Why would you submit a $1,000 non-refundable fee unless you liked
	flushing money down the toilet given this constraint and the likely
	response (hint: I bet 1,000 registration requests are filed within
	days of this process actually being opened)

2)	The process, process, process, process bent is still evident in
	Postel's latest draft.  135 days from first call for action to
	actually issuing any TLDs.  *135 days*, or FOUR MONTHS!  Why?

	Process for the sake of process, that's why.

3)	Again, the IANA wants to take control, to be "big brother", to
	insure that no customer gets screwed if a registry fails.  Well,
	that's very noble and all, but frankly, IANA and IETF folks, you're
	NUTS.  Do you want the liability that will inevitably follow?  What
	if someone gets a $250,000,000 judgement against SAIC and NSI and
	*THEY* fold?  Can you run that registry and servers?  Quite probably
	not if it happens early.

	Second, this 1% tax isn't being imposed on NSI.  Why not?  With
	400,000 registrations, and a $20 M annual revenue stream, I would
	think that $200,000 US annually would be very attractive to the
	IANA.

	If you have a right to tax the rest of us, you have the same right
	to tax NSI.  Or, is the truth still that you don't believe you
	actually own the top level space, but rather are a custodian of it?
	If that's true, then taxing ANY OF US is improper.

Let's take some truisms here and work from THOSE, rather than entrenching
the monopolists further and inviting one or more of us who have both lawyers
and money to go after one or all of the folks trying to fill the road with
rocks rather than solve the problems.

I'd like to see people attack these POSTULATES if you want to change the
principles that I based my draft on.  Go after these FIRST if you believe
there really is an issue here, because these are the truths which that
document was based on (and which the others are ignoring).

POSTULATES
==========

1)	The Top Level Zone (".") will NOT break with even a huge number of
	top level domains.  This has been operationally proven with the .COM
	zone, which has some 400,000 subdomains in it.  Therefore, we have
	proven that a zone (including ".") can handle up to 400,000
	delegations and NOT break down.  Further, .COM is doing this even
	while it is being housed on the same servers as hold the root zones
	-- a practice that nobody is seriously suggesting for real top level
	domains other than .COM (this practice just adds load to the top
	level root servers).

2)	Following from (1) above, The only *technical* reason to control top
	level delegations (ie: ".FOO", etc) is to insure that no two
	organizations each issue conflicting delegation announcements for it.
	That is, there can be only one authority for .COM, one for .FOO, etc.
	Likewise, there ARE real considerations for the current top level
	zones (such as the ISO two-letter country codes) which we must
	insure do not get stepped on.

3)	The current people who hold the International Registry TLDs (ORG,
	COM, NET) have been granted a monopoly without due process.  This
	has inured to their benefit a US $20,000,000 annual monopoly revenue
	stream (at the current 400,000 .COM registration benchmark.) The
	practices, business model, and books of this organization are not
	subject to public review.  It is known that they do not actually
	provide all of the network connectivity nor nameservers necessary to
	serve the zones they register names under; in fact, many of those
	servers are operated by other entites without apparent compensation.

4)	It requires little in the way of resources to actually operate a
	Top-Level Domain server.  Certainly, any multi-homed ISP (Internet
	Service Provider) could easily host such a thing in a reliable
	fashion.  The current *bandwidth requirement* to operate a root
	nameserver is under a single T1 on a peak-load basis.  These
	requirements are consistent with any mid-sized US ISP or larger
	organization, and are not beyond the reach of modest to moderate
	business interests.

5)	Under Postulates 1, 2 & 4, there is no *technical* justification for
	barring any organization meeting these requirements from operating a
	top-level domain server.  The Internet and its connectivity will
	*NOT* melt if these registration requests are all granted tomorrow.

6)	The only means to obtain a solution to the problem given by
	postulate (3) above is a free marketplace, with buyers and sellers
	able to deal with one another on a contractual basis without outside
	interferance.

7)	There exists no justification for any organization, including the
	IANA or IETF, trying to provide what amounts to "insurance" against
	a registry going bankrupt or otherwise ceasing to operate.  This is
	properly a contractual matter between registrants and the domain
	registries, with the risks present being properly allocated by the
	free market between these organizations.

	(Example: A "free" registry might have a high probability of going
	bankrupt.  Those firms and people registering under that top level
	domain would accept, as part of the fact that this service was at a
	very low cost or no charge, this risk.  Likewise, a firm charging
	$1,000 a year for registration services is unlikely to fail due to
	its inability to recover costs.  This risk is inherent in any
	business transaction and it is improper for the IETF or IANA to
	attempt to distort the free marketplace.  We have the S&L debacle in
	the US to consider; outside interferance as an attempt to prop up
	poor business decisions was partially responsible for the poor
	investment decisions these firms made, as they knew they would not
	be held accountable and the government would bail them out).

Following these postulates, I propose:

1)	That any registration authority (after: "registry") desiring to
	operate a top-level-domain service be registered on a first-come,
	first-served basis by the IETF and IANA.  The following rules shall
	prevail in the naming of these zones:
		Between three and SIX alphanumeric characters
		First character alphabetic
		Name of zone must be a "generic" and not generally
		recognized as a strong trademark (ie: you can't get ".IBM"
		as a top level domain)

2)	Disputes over precedence in registration shall be resolved by "first
	use" on the Internet in commercial activity, as with Trademark
	disputes generally.  Persons or firms unhappy with these results may
	take to the courts to fight amongst themselves, as always.

3)	The IANA, IETF, and ISOC may debate among themselves who holds the
	generally-recognized "top level zone file" to be distributed to the
	root nameservers.  Until otherwise resolved, this responsibility for
	inserting the records into this file shall lay with the IANA.

4)	Accepted applicants shall pay a US $1,000 one-time adminstrative
	fee to the IANA, to be distributed and spent in accordance with its
	charter.

5)	Applications may be made by postal or electronic mail to a
	designated address of the IANA's choosing.

If you wish to debate this, start with the POSTULATES.  "Because I don't
like it" is not a defensible position, folks.

Failing this, let's have at it with the registries.

--
--
Karl Denninger (karl@MCS.Net)| MCSNet - The Finest Internet Connectivity
Modem: [+1 312 803-3590]     | T1 from $600 monthly; speeds to DS-3 available
Voice: [+1 312 803-MCS1]     | 23 Chicagoland Prefixes, 13 ISDN, much more
Fax: [+1 312 248-9865]       | Email to "info@mcs.net" WWW: http://www.mcs.net/
ISDN - Get it here TODAY!    | Home of Chicago's only FULL Clarinet feed!


Richard Stiennon			richards@netrex.com
Director, Business Development		www.netrex.com/richard
Netrex, Inc.				Voice: 810-352-9643
3000 Town Center, Suite 1100		Fax: 810-352-2375
Southfield, MI 48075