Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

History of TCP/IP.

1 view
Skip to first unread message

Alan Robert Ford

unread,
Aug 16, 1997, 3:00:00 AM8/16/97
to

Hello there,

I was wondering if anyone out there knew of any URLs or other
sources that would tell me a bit about the history of the development of
TCP/IP and various implementations of it.

For example, the history of the BSD TCP/IP implementation would be
great to have a reference for.

Thanks in advance,

Alan Ford

c932...@tesla.newcastle.edu.au

John Nagle

unread,
Aug 16, 1997, 3:00:00 AM8/16/97
to

Read the RFCs or the standard textbooks on the subject.

The BSD implementation came along well after TCP was deployed.
It wasn't even the first UNIX implementation. 3COM at one time
sold a product called UNET, which was a TCP/IP implementation for
4.1BSD and UNIX System 7. After 4.2BSD came out with Mike Karels'
implementation of TCP, UNET was dropped. 3COM was out of TCP/IP
for years after that; only later did they get back into the field.

I had UNET running on VAXen, PDP 11/70s, and even Zylog 8000
machines in the very early 1980s, all at Ford Aerospace. The
Zylog 8000 machines may have been the first single-chip microprocessor
machines on the Internet.

John Nagle

Guy Harris

unread,
Aug 16, 1997, 3:00:00 AM8/16/97
to

John Nagle <na...@netcom.com> wrote:
>3COM at one time sold a product called UNET, which was a TCP/IP
>implementation for 4.1BSD and UNIX System 7.

(Well, "Version 7", or "Seventh Edition", but....)

It also included a mechanism for sending IP datagrams over serial lines.

Computer Consoles, Inc. was building an office automation system for the
Naval Surface Weapons Center (or whatever center it was) at Dahlgren,
VA; it used a couple of VAXes running a 4.1cBSD/4.2BSD tweaked to be
S3-compatible and a bunch of 68K-based CCI Power 5/20 machines running a
S3ish UNIX. They wanted the machines networked, so we added UNET to the
CCI machines.

Most of the CCI machines had only serial lines for networking, so Rick
Adams (yes, *that* Rick Adams) did an implementation of the 3Com serial
line encapsulation of IP for BSD so the VAXes could talk to the CCI
machines. It was implemented as a "line discipline", the #defines
giving the numbers of which ended with "DISC", so he called it "Serial
Line IP", or SLIP (yes, *that* SLIP), so that the #define could be
"SLIPDISC". He then sent "if_sl.c" back to Berkeley.

So, thanks to Rick Adams and Marc Andreesen, we have now have billboards
and TV ads and drain-cleaning service trucks and... with URLs. :-) I
suspect the existence of SLIP, and of its successor PPP, may have helped
bring the Internet to the masses; without that, and without a whizzo
browser capable of displaying flashy ads, we might not have had the Web
in the hands of the masses.

(And, no, he did this long before he had the wacky idea of setting up a
company to act as a giant UUCP hub, so he didn't do it to lay the
groundwork for UUNET....)
--
Reply, or follow up, but don't do both, please.

postmaster@localhost
postmaster@[127.0.0.1]

Hwa-Jin Bae

unread,
Aug 18, 1997, 3:00:00 AM8/18/97
to

na...@netcom.com (John Nagle) writes:

> The BSD implementation came along well after TCP was deployed.

>It wasn't even the first UNIX implementation. 3COM at one time


>sold a product called UNET, which was a TCP/IP implementation for

>4.1BSD and UNIX System 7. After 4.2BSD came out with Mike Karels'
>implementation of TCP, UNET was dropped. 3COM was out of TCP/IP
>for years after that; only later did they get back into the field.

The original TCP/IP code for 4.1 BSD running on VAX 11/780s (and 750s)
was from BBN. This base was further modified in 4.1c BSD with older
socket interface (incompatible in certain ways with 4.2 BSD) and then
later further worked on by Berkeley folks. At roughly about the same
time (in early 80's) I remember using TOPS-20 (and TENEX) TCP/IP.

> I had UNET running on VAXen, PDP 11/70s, and even Zylog 8000
>machines in the very early 1980s, all at Ford Aerospace. The
>Zylog 8000 machines may have been the first single-chip microprocessor
>machines on the Internet.

I believe the first microprocessor (not single chip though) machines
that got hooked into ARPANET were LSI-11 machines either running Dave
Mills' "Fuzzball" software, or version 6 UNIX with BBN TCP/IP.
At one point at least some of the Zylog 8000 machines ran UniSoft
version 7 UNIX with very basic TCP/IP support.

0 new messages