[news.sysadmin] Multiple Internet connections?

doering@kodak.UUCP (Paul F. Doering) (05/17/89)

Distribution: na
Organization: Kodak Research, Rochester NY

What is the common practice in connecting to Internet when a
commercial organization has geographically-dispersed research sites?
Does each research site connect independently to its closest
consortium (NYSERnet, NEARnet, etc)?  Is there a domain-name problem?
Can each of the company's research sites share the parent company's
domain-name, even if widely separated, as westcoast.poobarcorp.com,
eastcost.poobarcorp.com, sunbelt.poobarcorp.com...., even if joining
several consortia means that there is no common gateway machine?

Please reply or ask clarifiaction of the question by e-mail. Thanks.
-- 
  =========================     ======================================
   Paul Doering (for self)         Man will never arrive,              
   rochester!kodak!doering            man will be always on the way.   
  =========================     =============== -Carl Sandburg =======

david@ms.uky.edu (David Herron -- One of the vertebrae) (05/18/89)

Well, I tried mailing back to you but sending to 

	doering@kodak.UUCP (Paul F. Doering) 

resulted in the following bounce so everybody gets to read this
deathless prose while at the same time scratching their head over
why the Received: lines don't say it went to Korea while at the
same time the From: in the message says it *DID* come from Korea.


Received: from e.ms.uky.edu by s.ms.uky.edu id aa29535; 17 May 89 17:43 GMT
Received: from rutgers.edu by g.ms.uky.edu id aa07840; 17 May 89 17:43 GMT
Received: from rochester.UUCP by rutgers.edu (5.59/SMI4.0/RU1.1/3.04) with UUCP 
	id AA21061; Wed, 17 May 89 13:42:03 EDT
Received: by sol.cs.rochester.edu (3.2/m) id AA09350; Wed, 17 May 89 13:19:36 EDT
Received: from park.kp02.kodak.com by kodak.UUCP (3.2/SMI-3.2)
	id AA15462; Wed, 17 May 89 13:18:01 EDT
Received: by park.kp02.kodak.com (4.0/SMI-4.0)
	id AB08049; Wed, 17 May 89 13:17:55 EDT
Date: Wed, 17 May 89 13:17:55 EDT
From: Mail Delivery Subsystem <Mailer-Daemon@park.gsc.co.kr>
Subject: Returned mail: Host unknown
Message-Id: <8905171717.AB08049@park.kp02.kodak.com>
To: david@ms.uky.edu
MMDF-Warning:  Parse error in original version of preceding line at e.ms.uky.edu

   ----- Transcript of session follows -----
link access: Unrecognized node name
550 doering@icts01... Host unknown

   ----- Unsent message follows -----
Return-Path: <rochester!rutgers!ms.uky.edu!david>
Received: from kodak.UUCP by park.kp02.kodak.com (4.0/SMI-4.0)
	id AA08046; Wed, 17 May 89 13:17:55 EDT
Received: by kodak.UUCP (3.2/SMI-3.2)
	id AA15455; Wed, 17 May 89 13:17:53 EDT
Received: by sol.cs.rochester.edu (3.2/m) id AA07578; Wed, 17 May 89 12:43:56 EDT
Received: from ukma.UUCP by rutgers.edu (5.59/SMI4.0/RU1.1/3.04) with UUCP 
	id AA13308; Wed, 17 May 89 12:25:21 EDT
Received: from s.ms.uky.edu by g.ms.uky.edu id aa04712; 17 May 89 16:21 GMT
To: rochester!kodak!doering
Subject: Re: Multiple Internet connections?
Newsgroups: news.sysadmin
In-Reply-To: <1886@kodak.UUCP>
Organization: U of Kentucky, Mathematical Sciences
Cc: 
Date: Wed, 17 May 89 16:20:04 GMT
From: David Herron -- One of the vertebrae <rochester!rutgers!ms.uky.edu!david>
Sender: rochester!rutgers!ms.uky.edu!david
Message-Id:  <8905171620.aa20496@s.s.ms.uky.edu>

These are my opinions mostly ...  I don't know any Official Policies
but I suspect that news.sysadmin is a bad place since it's for administrators
of Usenet -- not of the Internet.


In article <1886@kodak.UUCP> you write:
>Distribution: na
>Organization: Kodak Research, Rochester NY
>
>What is the common practice in connecting to Internet when a
>commercial organization has geographically-dispersed research sites?

I know that lots of companies have their own internal networks
which they use for internal business.  I think it'd be rather unfriendly
to use NSFnet for internal business but I don't know of any Official
Policy which'd prohibit that.

>Does each research site connect independently to its closest
>consortium (NYSERnet, NEARnet, etc)?  

I know that DEC's machines are connected to Cypress, BARRnet and
some NSFnet in the DC area -- at least.  DECWRL is on Cypress & BARRnet
and the decuac machine is in the DC area.  I don't see anything stopping
you from joining more than one consortium.

On the other hand I know that Nysernet has members outside of NY State
and you might think that's outside their jurisdiction ...

>Is there a domain-name problem?
>Can each of the company's research sites share the parent company's
>domain-name, even if widely separated, as westcoast.poobarcorp.com,
>eastcost.poobarcorp.com, sunbelt.poobarcorp.com...., even if joining
>several consortia means that there is no common gateway machine?

No problem here at all!  You could, for instance, declare a kodak.com
gateway and for any sub-domains which you want explicit gateways for
declare them as you need... all with MX records or usenet entries
or whatever.


-- 
<- David Herron; an MMDF guy                              <david@ms.uky.edu>
<- ska: David le casse\*'      {rutgers,uunet}!ukma!david, david@UKMA.BITNET
<- By all accounts, Cyprus (or was it Crete?) was covered with trees at one time
<- 		-- Until they discovered Bronze
-- 
<- David Herron; an MMDF guy                              <david@ms.uky.edu>
<- ska: David le casse\*'      {rutgers,uunet}!ukma!david, david@UKMA.BITNET
<- By all accounts, Cyprus (or was it Crete?) was covered with trees at one time
<- 		-- Until they discovered Bronze

hoffman@pitt.UUCP (Bob Hoffman) (05/26/89)

Paul F. Doering of Kodak asks about having multiple research sites
connected to different parts of the Internet.

This can be done easily.  In order to make sense of this, it is necessary
to understand that numeric internet addresses and domain names are
completely separate things.  Each research facility would apply for
a separate internet network number.  The company as a whole would
apply for a domain name, and would have complete freedom to assign
names under that.  In general, the network number is related to the
physical connection of the network and the domain name is related to the
organizational or political hierarchy of the company.

It is possible to have two different workstations in the same domain
but in geographically separate areas.  Example name server entries:

$ORIGIN plastics.poobarcorp.com.
hosta	IN	A	193.001.001.025
hostb	IN	A	193.001.002.037

In this example, class C networks 193.001.001.x and 193.001.002.x
may be half a continent apart, but the routing tables maintained by
the Internet will allow packets to be routed properly to each of them.
The fact that they have names within the same subdomain of poobarcorp.com
has no relevance whatsoever.


It is also possible for a company to apply for a single class-B
or class-A network number, but it is that company's responsibility
to provide complete connectivity between all parts of that network
internally.  It is not possible to have different pieces of the same
network connected to the Internet in different places.


I have only addressed the technical issues here.  The NSFnet or
regional network administrations probably have rules applying to
this situation.


-- 
Bob Hoffman, N3CVL       {allegra, bellcore, cadre, idis, psuvax1}!pitt!hoffman
Pitt Computer Science    hoffman@vax.cs.pittsburgh.edu