[1167] in Commercialization & Privatization of the Internet
Who wasn't at IETF.
daemon@ATHENA.MIT.EDU (Jack Haverty)
Tue Aug 13 01:44:37 1991
Date: Mon, 12 Aug 91 22:40:58 PDT
From: Jack Haverty <jhaverty@us.oracle.com>
To: Gene.Hastings@boole.ece.cmu.edu
Cc: ari@mordor.stanford.edu, com-priv@psi.com, becker@nis.ans.net
In-Reply-To: Gene.Hastings@boole.ece.cmu.edu's message of Saturday, 10 August 1991 23:35:42 EDT <1991.8.11.3.26.0.Gene.Hastings@boole>
Gene et al,
Good question:
"...
Is there any network operator on this list who does not participate in IETF?
In FARNET? Why?
..."
I'll take a crack at answering.
The easy answer is "no time". IETF participation requires a significant
commitment of time, not only to go to a month (!!) of meetings a year, but also
to prepare for such meetings and act on results.
But on thinking about this, I think there's something else going on. Oracle is
connected to "The Internet", but our primary focus is operating and managing a
world-wide internet which supports the Corporation. So we're heaviliy involved
in operating AN internet, but only peripherally connected with the operation of
"THE Internet".
My perception of the work being pursued for THE Internet is that it is very
forward-looking, and focusses around two primary goals - providing higher
throughput and interconnecting tens of thousands of sites.
For AN internet, in a commercial environment, these are not necessarily the
primary goals. In fact, unless you're a PTT or other carrier, interconnecting
hundreds of sites is adequate. Unless you're heavily into imagery (e.g.,
Kodak), providing throughputs of kilobits or maybe a megabit is the target -
both for reasons of costs and, in the international context, because
high-bandwidth services are simply not available.
Also, the users are different: there are lots of LANs and workstations scattered
around, but also lots of async terminals, 3270s, et al; there is electronic
mail, file transfer, and client/server traffic, but also lots of echoplex
terminal traffic; there is TCP/IP, but also DECNET, IPX, APPC, CLNP, XNS, and
other such stuff floating around.
And lastly, costs vary widely. In the US, circuits are relatively inexpensive;
in Europe they're very costly; in some areas only PDNs are available, and they
come with rather impressive usage-based charges - e.g., order of $.50 per
kilobyte.
So, the task of network operations in this context is to create a network which
provides extremely reliable service, deals with both the new user technology and
the current installed base, can be operated by technicians who didn't invent the
protocols and don't need 2 years of OJT, and uses communications services around
the world in a cost-effective manner.
So, the long answer to your question is that I think the IETF is working on
interesting areas that will advance the state-of-the-art in both throughput and
sheer size -- but we're struggling with other requirements that don't appear to
be on the IETF agenda (or on the vendors' agendas for that matter - but that's
another topic someday).
'nuff for now...
Jack
