[prev in list] [next in list] [prev in thread] [next in thread] 

List:       namedroppers
Subject:    Re: Domain Name System TLDs (fwd)
From:       "Donald E. Eastlake 3rd (Beast)" <dee () skidrow ! lkg ! dec ! com>
Date:       1993-12-18 16:11:17
[Download RAW message or body]


From:  Hank Nussbacher <HANK@VM.TAU.AC.IL>
To:  emv@garnet.msen.com, Hank Nussbacher <HANK@vm.biu.ac.il>
Cc:  emv@garnet.msen.com, William Manning <bmanning@is.rice.edu>,
            Jon Postel <postel@isi.edu>, namedroppers@nic.ddn.mil, qed@rice.edu
>I have a feeling that what we are cooking up here is gonna have the end
>users up in arms.  The teacher in Oshkosh who finally remembered that mail
>from .FR is from someone in France will have to throw out that little bit of

Anyone who thought that mail from .xx meant mail from a *person* in
.xx was always wrong.  Anyone who thought that mail from .xx meant
mail from a person who happened to have an account on a *host* in .xx
was likely to be right much of the time.  Anyone who thought that mail
from .xx meant mail from a person who happened to have an account on
a host that had *something* to do with .xx was exactly right.  The
host had enough associaton with .xx to bother registering in the .xx
domain.

>learned network oral bible and ask where the person is from.

>What about all those AI readers that scan the headers announcing who
>the mail is from and from where?  Chuck them cuz .AU could be Poland.

That's right.  They were always wrong.  And will be even more so with
mobile hosts.  If someone ads a latitude and longitude header to their
mail and you trust it, you can say where the mail is from.  Otherwise
all bets are off.

>Imagine two people in Germany on the net.  One is in the Brazilian DNS
>the other in the Australian DNS.  Imagine the wasted international
>bandwidth to do the lookups (please don't lecture about caches and
>T3 access, since I am looking to tomorrow when the Internet will have
>hundreds of millions of users - based on the 80% current growth we will
>close to 38 million Internet nodes in 5 years) in order to send a piece
>of email over a 2 hop ISDN local connection.

First of all, your argument is completely wrong in relation to
*people*.

Second, if you are refering to two *hosts*, then there will be cases
where some few UDP datagrams have to go a long way, but I really don't
think it is of any significance.  If the hosts are mobile, some amount
of communication with their home is needed under most schemes.  If the
hosts are fixed, you have a tiny scrap of ground to stand on, but not
much.  In the longer run, an isolated single host will be the very
rare exception.  The norm would be a business office or home or
factory or vehicle or etc. with multiple hosts which will be its own
DNS zone.  I know of at least two households that have their own
network number, have mutliple hosts, and are their own second level
DNS zone.  It makes very little difference if such places are
registered under .com or .us (where they happen to be) or .xx for some
other country or .org or whatever.  The primary DNS server for them
is on their premises and secondaries can be wherever its useful.

>Boy are we gonna look stupid when Data Communciations or Communications
>Week gets a wind of this.

Boy are we going to look fascist if we try to enfore rules of geopolitical
purity on DNS registrars who have nominally been deligated the right to
register whatever they want.

>Hank

Donald

[prev in list] [next in list] [prev in thread] [next in thread]