[32] in Commercialization & Privatization of the Internet
re: Was a "big Internet" needed to make TCP/IP useful?
daemon@ATHENA.MIT.EDU (Bill Nowicki)
Thu Oct 25 14:10:57 1990
Date: Thu, 25 Oct 90 10:17:42 PDT
From: nowicki@Legato.COM (Bill Nowicki)
To: craig@NNSC.NSF.NET, gnu@toad.com
Cc: com-priv@psi.com
Subject: re: Was a "big Internet" needed to make TCP/IP useful?
From: Craig Partridge <craig@NNSC.NSF.NET>
Date: Tue, 23 Oct 90 08:44:35 -0400
As I recall, DARPA alone was responsible for over 10% of Sun's
sales in the early years. Those machines were going on the
Internet. Add purchases from Comp. Sci. depts and labs that
wanted Internet-ready machines and you've got a big market
share.
I fought many battles when I was at Sun trying to get the management
to notice the Internet. Their attitude was that even though
Universities might be a significant chunk of revenues, the discount
rate was fairly high, so the amount of profit was quite low. Their
estimate was about 10% of Suns were on the Internet, but my estimate
was more like about 50%-75%. Unfortunately I could never get anyone
interested enough to find out any real numbers!
We are getting a bit off the point here. I have to agree with
John: having the government artificially speed up Internet growth is a
BAD thing. One example: Sun provided very little support for DNS, MX
records, routing, etc., because customers typically just FTP'ed the
sources from some other university and re-invented the wheel instead
of insisting on the vendor fixing the problem. If the true cost of
naming and routing were apparent, it would have been more likely that
they would have gotten some attention.
More important is to take a look at where the real innovations in
networking have come. The pioneers like Nagle, Karn, and Jacobson all
had to get clever and do things BETTER instead of using brute force.
If we continue to subsidize innefficiency, there is little motivation
for innovation (just look at the US automobile industry).
Every IETF meeting we hear about how the links have only 1-3%
utilization, but performance is bad, so the goverment is going to
speed up the links. The role of the government should be to fund
research to find out WHY performance is so bad at such low
utilizations. The Internet is already growing so fast that nobody
really understands how it works -- the last thing we need is to make
it even larger and less efficient.
The NSF should cancel the 45 Mbps upgrade and apply the money to real
research instead. They keep saying that the current upgrade is just a
stopgap, but I have been hearing that for many years. The longer it
continues, the worse the inevitable transition will be.
- WIN
