NAVYJOBS.COM

Vadim Antonov avg at postman.ncube.com
Mon Apr 22 10:07:25 UTC 1996


The current problem with DNS is not load on root NSes
(however, even a blind person can see how exponential
growth at more-than Moore's law will get it to the
point where you simply won't be able to find a box
fast enough to run root NS). Those of us with longer
memories remember times when 16Mb on AGSes was "more
than enough for any foreseeable future".  And, yes,
followinmg your analogy, 99.999% of routers do not have
to have more than 100 routes in them.  Does it help
a dozen of dying backbone boxes much?

The current problem is increased name collisions.  Which is,
to say, a fundamental one.

No matter how you jump you won't be able to decrease rate
of name collisions w/o increasing depth of the tree.
The only other remedy is to stop choosing natural
abbreviations and start doing as AOL does with customer
names.  Which is, essentially, an inferior way to do
exactly what i proposed to do.

(And, BTW, there's life after 26 letters are done with;
you can set moving "line in the sand" on who gets shorter
names).

But -- the idea that first level domains are sexy got
to be stomped out before anything real can be done about
DNS.

New TLDs is merely moving the problem one level higher.
It definitely cannot fix anything.

--vadim

PS.	No, i don't call to "rename Internet".  I call to
	stop the insane practice of building the flat
	namespace.

>From GAVRON at ACES.COM Mon Apr 22 17:45 PDT 1996
Return-Path: <GAVRON at ACES.COM>
Received: from postman.ncube.com by butler.ncube.com (5.0/SMI-SVR4)
	id AA16312; Mon, 22 Apr 1996 17:45:36 +0800
Received: from Hearts.ACES.COM by postman.ncube.com (4.1/SMI-4.1)
	id AA18743; Mon, 22 Apr 96 17:47:04 PDT
Received: from ACES.COM by ACES.COM (PMDF V4.3-10 #4107)
 id <01I3RYXO7X0G000031 at ACES.COM>; Mon, 22 Apr 1996 17:46:44 -0700 (MST)
Date: Mon, 22 Apr 1996 17:42:30 -0700 (MST)
From: Ehud Gavron <GAVRON at ACES.COM>
Subject: Re: NAVYJOBS.COM
In-Reply-To: Your message dated "Mon, 22 Apr 1996 15:45:23 +0800"
 <9604222245.AA15747 at butler.ncube.com>
To: avg at postman.ncube.com
Cc: john at interport.net, smd at icp.net, nanog at merit.edu, GAVRON at ACES.COM
Message-Id: <01I3UN0PL9NS000031 at ACES.COM>
Organization: ACES Research Inc.
Mime-Version: 1.0
Content-Transfer-Encoding: 7BIT
Content-Type: TEXT/PLAIN; CHARSET=US-ASCII
Content-Length: 1809
Status: R


Vadim asks:

>Why not to restrict first-level domains to companies
>which can demonstrate that they have 1000+ hosts?

Creating a problem to solve a problem is not a solution.
Paul already said that a 16MB 486 can handle root DNS just fine.
This does not bespeak of a problem requiring "renumbering"
hundreds of thousands of domains.

FURTHER, your "big win" is only a 1/26 lessening of load.
If the problem is such that 1/26 is a "big win" I think
it's not big enough to rename the Internet for.

...
>What we should worry about is number of first-level
>domains/number of hosts ratio.  It is the same problem
>as with routing.  The solution is also the same --

No, it's not the same problem as routing.  All routers with
non-default rules need to know about all routes.  The only
possible analogy for that in DNS is root nameservers.  These
(as Paul pointed out in a previous post, since he runs F)
are not saturated, and don't require that many resources.

If you feel the growth of domain names is such that it will
outstrip a 486 w/16MB soon, tell me when it will be a
SIGNIFICANT problem.  I.e. when will it outstrip a real
machine (Sun, VAX, Alpha, SGI) with real memory (64MB?
128MB?  4G?) 

Remember, upgrading HUNDREDS of routers all running 45Mbps
is a priority.  Upgrading 8 boxes running BIND and doing
it well is a much much much lower priority.


>mnemonic can be used to distinguish between thousands of
>nearly identical small businesses?)

>--vadim

>PS.  Obviously if IBM registers 100 domain names it is
>     still a lot less damage than a small ISP (with 1000 dial-up
>     customers) which registers a domain name for every such
>     customer.  Big folks registering POISONOUS-BURGER.COM and
>     SHIT-ON-TV.COM aren't really a problem.  Zillion of
>     MOM-AND-POP.COMs is.

Ehud




More information about the NANOG mailing list