In-Reply-To: <199504061303.AA10477@world.std.com>
Message-ID: 
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII



On Thu, 6 Apr 1995, Daniel P Dern wrote:

> 
> Rex Ballard says:
> 
> > Remember, until 1991 the internet was a bunch of modems, UNIX (4.2 BSD yet)
> > and ethernet cards that could "sniff" TCP/IP packets. 
> 
> Funny, back when I started working at Bolt Beranek and Newman (BBN) back in
> 1983, I could have sworn that over in the Network Operating Center (NOC),
> there was a big map labelled "Internet" with things like a 10 megabit per
> second chunk running through it labelled WIDEBAND, something else labelled
> SATNET, among other things... not to mention a whole lotta non-UNIX
> hosts.  I also wonder whether there was Ethernet cards; mostly there
> were hosts connected to IMPs (packet switches acting as routers, aka 
> gateways or vice versa).

I was assisting on the Dalgrin Project in 1982.  There were LOTS of TCP/IP
networks, liked together with everything from X.25 and modems to ethernet 
cable strung between buildings and between floors of office buildings.  
Boulder Colorado used to include internet links in their incubators.

BBN was one of several LAN/WAN vendors.  One of the first commercial 
"Public Access" boards.  The backbones consisted of DARPA, HP, DEC, and
Texas Instruments.  There were several points of convergence, one was
in Colorado Springs, Colorado.  In most areas, you would get an internet 
account through the local university in exchange for contributing free 
source code, technical services, or other quid-pro-quo.  Folks in Boston 
used to Brag about the quality of BBN service.  The rest of us had to
squirm through slow links to the Hacker's University :-).

> And what was the NSFNET that acted as the de facto non-commercial backbone
> starting in 1989... or PSInet, UUnet, NEARnet, etc -- chopped liver? These
> had T1 and greater lines from the word go; if you want to call T1 speed routers
> and CSU/DSUs or whatever "modems" you can, but I think it's disingenuous
> to call them modems.  And technically inaccurate.

NSF was prohibited from connecting directly to most businesses unless 
they provided research and information to the NSF.  Using NSF backbones 
for commercial TCP/IP traffic prior to 1990 was verboten.  It was amazing 
what HP and SUN could pass off as "research".

> ALso, let's not forget that the first TCP/IP wasn't done for BSD Unix, it was
> done for PDP-11 Unix, then later ported.  (p.11, The Internet Guide for New
> Users, by yours truly, quoting Jack Haverty, the guy who did it.)
Yes, it was the first GPL software, Version 6 Unix.  AT&T couldn't sell 
it so they gave it to UCB, CMU, and MIT, who promptly went bananas with it.
Every MIT student got a notebook of the annotated UNIX source code.

> Your comments re Cliff Stoll's book have merit.  But I wonder where you
> got the above facts from, they don't fit in with my direct and indirect
> experience.

There are some things we don't generally admit to, like using NSF 
backbones for commercial traffic.  The barriers and restrictions weren't 
lifted until 1990, and commercial "Supernets" such as BBN, WestNet, and 
Colorado Supernet were not "self supporting" until early 1992.  The 
"offices" were closets in the back of a few universities.  Today, MCI, 
AT&T, SPRINT, and several Cable companies carry Internet traffic over 
multple 1.5 gigabit DS5 links.  That's a lot of growth in less than 3 years.

> DPD

> Daniel Dern (ddern@world.std.com) Internet analyst, writer, pundit & curmudgeon

	Rex Ballard


From rballard@cnj.digex.net Fri Apr  7 10:10:53 1995
Status: O
X-Status: