[1648] in Commercialization & Privatization of the Internet
Re: Commercial Confusion
daemon@ATHENA.MIT.EDU (Erik E. Fair" (Your Friendly Postm)
Mon Dec 9 06:36:36 1991
From: "Erik E. Fair" (Your Friendly Postmaster) <fair@apple.com>
In-Reply-To: <9112081509.AA09701@tmn.com>
To: cook@tmn.com (Gordon Cook)
Cc: com-priv@psi.com
Date: Mon, 09 Dec 91 03:35:49 -0800
Date: 8 Dec 91 15:09:37 EST (Sun)
From: cook@tmn.com (Gordon Cook)
Subject: Commercial Confusion
It is likely that MOST Fortune 1000 companies priorities are
FIRST for widearea TCP/IP interconnects of their LANs and only
secondarily for talking to the colleges and universities on the
NSFnet. Faced with this jungle of barriers as to what is
acceptable and what is not why wouldn't they opt for MCI's
private widearea tcp/ip interconnects having nothing to do with
the internet community? Or for EINet?
This point of Gordon's is not very clearly expressed, and I think that
I can offer some information by way of clarification. Apple Computer
(and I think we're somewhere in the Fortune 1000 still) is on the
Internet for inter-organizational communication, not intra-organizational
communication. Put simply, we do not run our internal traffic over the
NSFNET (or for that matter, CIX) backbone. In my professional opinion,
the technology to run our internal traffic over the Internet in a
secure manner with confidence that it would not be inadvertently
exposed is not yet deployable (presuming that it exists). This is why
we use private leased circuits for our long haul network (of course, we
also have a strong need to run a proprietary networking protocol along
side TCP/IP: AppleTalk).
Secondly, so far as I can tell from the records I have here (Apple
connected to the Internet before they employed me), the initiative to
connect to the Internet (specifically, CSNET) in the first place came
from Apple's Advanced Technology Group (ATG - the long term research
organization), with the desire of collaborating with researchers
elsewhere. Higher Education marketing also has a strong interest in the
Internet, since a large fraction of their customers are on the Internet
(they provided the initial impetus for the AppleLink/Internet E-mail
gateway), but operational and administrative control of Apple's
Internet links is firmly in the hands of the Engineering department.
Now, having said all of that, I am still quite interested in seeing an
Internet that is open to commercial traffic, on a number of grounds. I
think that we have information services to make available to both the
R&E and commercial sites that are difficult to justify in the current
confused environment, and certainly I would like to run Apple's
internal and inter-company traffic over the Internet, if I can be
convinced that we can do it in a secure manner. Just to name two
reasons.
One other item: William Schrader's message stated that the T3 backbone
was not yet needed. I won't directly disagree with that statement; I
will offer a short allegory instead. Apple's Engineering Network
stretches from Tokyo to Paris, with over 100 IP subnets, and 600+
AppleTalk nets. A year and a half ago, our backbone was a combination
of 50Mbit/sec NSC HYPERchannel (for IP), and Ethernet (for AppleTalk).
We bought and installed an FDDI ring for our four main engineering
buildings at a time when the main ethernet wire was not at or near
capacity (or even half capacity, I think, but I'd have to check that).
Within three months, the new FDDI backbone was carrying more than
10Mbit/sec sustained traffic throughout normal business hours (i.e. we
passed the point of no return). Our FDDI ring still has exactly one
host on it: our Cray Y/MP-2E. All other FDDI attachments are network
routers to Ethernet, so the most any single host (other than the Cray)
can blast at the ring is 10Mbit/sec. The only way we get more than
10Mbit/sec on the ring all day long is aggregation. But boy do we
aggregate...
Erik E. Fair apple!fair fair@apple.com
