Linux and growth of Internet

From: Steven M Jones <classiccmp_at_crash.com>
Date: Thu Jan 13 12:20:12 2005

Sellam Ismail wrote:
.
> I can make a really good argument ... that e-mail could never have
> gotten as big ... as it did without the modern (1990+) Internet,
> which could not have gotten where it is today with Linux.

I assume that was supposed to be "without Linux." I'll bet someone
like Tom Jennings will have a totally different take on this than I
do, too. [ Ah, there's his note... Clearly I've been editing this
response too long. ]

Where to start... I disagree about the growth of the Internet not
being possible without Linux. It would have been different, perhaps,
but I honestly don't think it would have changed enough to surprise
someone who popped into a Linux-less parallel universe as regards
email and the Internet. Might have changed a lot more regarding the
popularity of open source software, but that's a different
discussion.

I think that the growth of email and the Internet in everyday life
rode two big trends. One is the rise of CompuServe and AOL, which
I'll leave for another discussion. The second was the rise of small
scale ISPs in the 1993-1994 timeframe, at least in the Northeastern
US, and that they were offering raw dialup IP connections. They'd
give you a POP account, maybe shell access, and that set the stage
to a great degree for non-corporate, non-academic participation in
the Web when it came on-stage.

Regular folk with Windows/Mac machines were signing up for these
accounts, not just tech weenies, for all the reasons that email was
popular. No long distance charges, asynchronous between sender and
receiver, faster than the post, etc. I recall some who were
downloading freeware and shareware like crazy, but the major thrust
for this non-tech segment I observed appeared to be email, with some
IRC/chat junkies. This led a gentle but steady ramp that was ready
to explode with the advent of Web browsers. And the low-usage email
customers were what supported the model, not the high-usage geeks
who would try to nail up connections all the time for the low-ball
prices.

Why don't I think Linux was a major factor in this? Surely many of
these ISPs used Linux? Well, maybe so. The ones I dealt with would
use whatever they could get their hands on to provide the backend
services they needed, and didn't care what they had to use to do
it. No Linux? Hey, that free thing 386BSD works, or this 1993 fork
called NetBSD. But to be honest, most of the time they very, very
grudgingly spent the money for Sun gear, at least for the most
critical machines/services. And then ran them without any service
contract or spares on-hand, which always made me cringe... At
least, this is what I ran across in my adventures, your mileage
obviously varied.

But again, I wouldn't give Sun credit for the spread of the 'Net and
email either, for the same reason. If it hadn't been Sun, it would
have been something else, and I ran into odd bits of DEC and SGI
here and there. I think there were several factors coming to bear,
like the NSF AUP changes, like the appearance of PCs with IP stakcs
in many more homes, like falling prices for the low-end leased lines
that the new ISPs depended would massively oversubscribe... ;^)

You want a technical change to point to? I'd point to TCP/IP coming
to Windows and the Mac. (Yes it had started much earlier, but it was
now getting easier and more reliable.) I'd point to terminal servers
getting good SLIP/PPP and connection accounting support. And
whatever it was that prompted the Baby Bells to start providing DS0
lines and fractional T1 loops to anybody who could scratch up the
cash, if that was in fact a technical change.

Would many companies have done things differently without Linux?
Sure. But even so, I think it would have happened at a slightly
reduced pace with *BSD under the cloud of the USL/UCB lawsuit and
the high price of a BSDI license -- which when I looked at it was
closer to $1,000 per host, not $100 for a site license! (Might have
been a built-in support cost, memory is fuzzy...) The hardware
savings alone would have continued to make that a compelling
combination versus $10-20k for Sun workstation.


Okay, I've spouted off enough for now. I'm curious to see what other
responses come in, and what others remember from that time, now as
on-topic as anything else since it's more than 10 years ago. Gulp!

--Steve.
Received on Thu Jan 13 2005 - 12:20:12 GMT