Re: .COM Clusters are Not RSCs

[Karl Denninger <karl at mcs.net>]

On Fri, Mar 20, 1998 at 03:42:25PM -0500, Keith Moore wrote:
> > > That addresses only one of several concerns.  It presumes that someone
> > > has the right to "own" a particular gTLD, if he can sign up some
> > > number of users.  I disagree with that presumption.
> > 
> > Then go argue with every TLD operator currently in existence, including .US.
> 
> Read my statement again.  (hint: you missed the 'g'.)

You misLED people by insisting on claiming that there is a difference.

THERE IS NOT.  Several so-called country-code TLDs are being farmed out and
registrations sold to any and all comers, REGARDLESS of the nationality of
the registrant.

A TLD is a TLD.  Deal with reality Keith, or shut down .TO and others which
have turned into completely-commercial havens.

> > There is not ONE current TLD that has this policy in the root.  
> 
> That's because the folks who were paid to administer those TLDs have
> been acting like they own them.  That's the root of this whole mess.

Gee, and the people who were given them for free (ie: xxxx.IL.US from .US)?  
They aren't doing the same thing?  They certainly are.

> > I believe you and others with this point of view are simply tilting
> > at windmills with this position, as the precedent is solidly on the
> > other side of the equation.
> 
> Your mind is truly twisted.  Somehow the root of this whole problem
> becomes a 'precedent' that is used to justify perpetuation of the
> problem.

Again, there are ~200 precedents and ALL OF THEM speak to the same point -
that TLDs are, in fact, owned.  

You're trying to paint NSI as the "bad guy".  Unfortunately, the REAL problem
lies squarely within the IANA and Postel's cranium, in that he thinks he can 
play tin-pot dictator and God in handing out monopolies ONLY TO A SELECT FEW 
PEOPLE, and in fact he HAS been doing this since he got that responsibility.

The entire REASON that COM is valuable is that there are few other choices!

Destroy THAT, and the profit motive disappears.

CORE/POC *ENHANCES* that monopoly position.  By doing so, we WILL see higher 
fees - which are completely unjustified.

Don't believe me?  Go look at some of the CORE people's pricing.  I've seen,
personally, $95/year on one.  And you thought $50 was a rip-off?  Get ready 
to bend over and grab your ankles, because you haven't seen anything yet.

I personally believe that we can run a couple of TLDs at $15/year/name
and make a tidy profit.

Now which would you prefer to pay:

1)	$15/year
2)	$95/year

For the SAME SERVICE.

If we do not have an open root, you will NOT get the first choice.  You WILL
get the second, and I know this because it is ALREADY BEING OFFERED TO THE
PUBLIC by people claiming that POC/CORE will get their TLDs in the root!

I note for your edification that CORE is going to charge, according to the
last mail from their list that I saw, $15/year/domain FOR THE BACK END
DATABASE SERVICES *ALONE*.

I believe we can do *THE WHOLE THING*, including the front office functions,
for that price - and not take a loss - even though I'm doing to do SEVERAL
times as much work as CORE is for the same amount of money.

CORE is a *HUGE* rip-off of the Internet user.  

That much, at this point in time, is established.

> > A query for X.COM has to go to the root server, and THEN to the server which
> > holds the data for X.COM.  The roots run with recursion off.
> 
> I stand corrected.  But it's still the case that a query for X.Y
> requires one more round-trip (assuming current software and large
> numbers of Y) than a query for present-day X.COM.

Yep - one more round-trip to servers running with 1/100th of the PRESENT load!

> > This means that the funnel function is FLAT; there is ZERO cache coherency,
> > since the roots take ALL the hits for the names in the zone (since they hold
> > the zone) and then refer the hits out to the server with the "A" record (or
> > MX, or whatever).
> 
> That's not true for all the roots.  Some of the roots currently also
> serve some of the gTLDS, but not all of them.  And this is a
> configuration issue, not an inherent limitation or feature of the
> current protocols or software.

But the argument you're making is that the root will BREAK if we load a
bunch of TLDs.  That's false, and our operational experience PROVES that 
it is false.  There is simply no intelligent debate possible on this point,
because we have multiple years of experience in this regard.

> > > If there were 15,000 gTLDs, chances are that a large number of their
> > > NS RRs would not be cached.  A query for X.Y would thus require
> > > additional round-trips.  Each round trip consumes more network
> > > bandwidth, increases the delay seen by the user, and increases the
> > > opportunity for failure.  
> > 
> > You are correct, but you proceed from an incorrect assumption.
> > 
> > Having lots more TLDs means that there MAY be a cache miss on the TLD lookup
> > itself.  But that's not really the issue.  The issue is the lookup for the
> > SLD within the TLD.  
> 
> That's not the issue I'm concerned about.

The issue you're concerned about ends up being a 90%+ net *DECREASE* in 
traffic to the root servers, not an increase.

This is a mathematical FACT, and I'm STILL giving you a 10x (one full order
of magnitude) less cache coherency - a figure that I consider rediculously
on the pessimistic side.

> > Right now there are a dozen machines that take ALL of
> > that load for the commercial domains.
> 
> So?  Machines are a tiny fraction of the cost.  If they need more
> machines, they'll get more machines, and they won't have any trouble
> paying for them.  You don't really want many more machines than
> necessary, because more machines means more failures and higher
> maintenance costs.  More machines also implies a wider variation in
> service quality (some machines, some software, some system
> administrators, some power sources, etc. will be flakier than others)
> which, after a point, translates into *less* reliability and makes it
> harder to diagnose and correct failures.
> 
> You obvously don't know beans about the scaling properties of
> distributed systems.  Try doing your homework next time.
> 
> Keith

On the contrary.  

You are arguing that the root systems will break if we allow 15,000 TLDs in
them tomorrow.  That's pure, unadulterated horseshit, and if you spend even
15 minutes of analysis on this with an understanding of how DNS actually
works, you'd know this.

Since you continue to propound that it will break, I conclude one of two
things:

1)	You don't know what you're talking about.
2)	You DO know what you're talking about, and you are DELIBERATELY
	MISLEADING PEOPLE.  The next question is "why?"

--
-- 
Karl Denninger (karl at MCS.Net)| MCSNet - Serving Chicagoland and Wisconsin
http://www.mcs.net/          | T1's from $600 monthly / All Lines K56Flex/DOV
			     | NEW! Corporate ISDN Prices dropped by up to 50%!
Voice: [+1 312 803-MCS1 x219]| EXCLUSIVE NEW FEATURE ON ALL PERSONAL ACCOUNTS
Fax:   [+1 312 803-4929]     | *SPAMBLOCK* Technology now included at no cost