Friendly Balkanization....

mo@bellcore.com Fri, 18 December 1992 15:10 UTC

Received: from ietf.nri.reston.va.us by IETF.CNRI.Reston.VA.US id aa05202; 18 Dec 92 10:10 EST
Received: from CNRI.RESTON.VA.US by IETF.CNRI.Reston.VA.US id aa05196; 18 Dec 92 10:10 EST
Received: from babyoil.ftp.com by CNRI.Reston.VA.US id aa12400; 18 Dec 92 10:12 EST
Received: from mailee.bellcore.com by ftp.com with SMTP id AA06188; Fri, 18 Dec 92 10:05:20 -0500
Received: from gizmo.bellcore.com by mailee.bellcore.com (5.61/1.34) id AA05847; Fri, 18 Dec 92 10:04:52 -0500
Message-Id: <9212181504.AA05847@mailee.bellcore.com>
To: kasten@ftp.com
Cc: tli@cisco.com, criteria@ftp.com
Subject: Friendly Balkanization....
Date: Fri, 18 Dec 1992 10:04:50 -0500
Sender: ietf-archive-request@IETF.CNRI.Reston.VA.US
From: mo@bellcore.com

Actually, the "friendly balkanization" that you describe (multiple nets
with applications gateways) has been the case for quite some time in
specific cases.  To wit:

The EMail nets have almost always been multi-protocol and mail
application gateways have been the norm since at least the mid-late
70s.  The most visible clumps in that galaxy include the fully-connected
SMTP cloud, the UUCP cloud, the FIDONET cloud, the BITnet cloud, the
commercial Email providers clouds, and others more obscure but just as
connected.  In general, unless one pings a nameserver, one seldom
actually knows what vehicles will be involved in delivering a given
piece of mail.  The Internet DNS does a great service far beyond its
direct reach.

For "news" applications, USENET and FIDONET are both in there, as are
others.

Most of these are cross-connected to an amazing degree, and yes, header
munging is awful and unclean and should be avoided at all costs, but it
works to an equally amazing degree.

Now that there are servers which respond to ftp requests sent via
EMail, the applications one migh class as "bit-vending" are working
trans-protocol universe.  Other gateways (FTP/NFS, FTP/FTAM, etc, etc)
are happening as well.

Admittedly this level of interconnection works best for services where
latency can be tolerated, but the Internet Gopher work is busily sewing
together lots of interactive but previously-disjoint services under a
common service delivery protocol and interface.  Adding arbitrary
gateways is relatively easy, and gateways exist between Gopher and
TELNET-based services,  Gopher and FTP, Gopher and WAIS, Gopher and
ARCHIE, Gopher and News, and as soon as a service pops up in some other
alternate universe of sufficient interest, someone will build the
gateway.

Note that I'm not claiming any of this is perfect, and given the
choice, one would prefer ubiquitous direct interoperability.  While it
is a grand goal, and one which we should vigorously pursue, it is
unattainable any time soon, and given that, the growing web of
applications gateways, bridging the transport and semantic gaps between
various network fabrics is indeed doing a great service bringing us all
together.


	-Mike O'Dell

Bellcore?? Bellcore isn't allowed opinions.  Any found here
are merely my mad ravings.