[ih] TCP adoption in 1984

[Steve Crocker]

>From 1981-85 I was director of the Computer Science Research Laboratory at
The Aerospace Corporation.  We connected the lab to the Arpanet using a
port on the Space Division IMP located across the street at the AIR Force's
Space Division headquarters.  A pipe underneath the street allowed us to
run a cable from their location to ours.  I don't recall which interface we
used; I assume it was a Distant Host interface—more than 50 feet but not
requiring a modem.  Once our lab was on the Arpanet, discussions commenced
about how to extend that access to other parts of Aerospace.  We settled on
a "ring with tails" architecture.  We installed a Proteon ring connecting
the various buildings.  The physical pathways were pipes that all led to
the main computer center building, A3, so the "ring" looked more like a
star, but it was topologically a ring.  We connected a PDP-11/23 to the
ring in each building and ran a 3MB Ethernet cable from it down the raceway
in the ceiling of the long corridor at the center of each building.  It was
thus possible to connect any computer on the main campus to this network
and gain access to the Arpanet.  This low-cost design could be replaced in
a few years when better technology becomes available without concern.

We had difficulty choosing our network name.  "Aerospace" was the natural
name, but in those days, it was common practice to have a much shorter
alternative name.  In formal settings, Aerospace made a point of referring
to itself as The Aerospace Corporation to distnguish itself from the myriad
of other aerospace companies, e.g. Ford Aerospace.  We considered using the
initials TAC, but that would have caused confusion with the Terminal Access
Concentrators (TACs) that replaced the TIPs.  The Aerospace logo is a
slanted capital A inside of a circle, and we briefly considered using "@"
as the short name.  Since that was the quote character on the TIPs and
TACs, a user trying to connect to Aerospace would have had to type two at
signs, "C @@" and would seen three because of the echoing.  And no one else
had names with special characters.  We discarded the idea after a few
moments of amusement.  Lacking a better idea, we settled on the single
letter A, which was suitably entered into the host table.  CMU promptly
stopped working.  Their machines were named CMU-A, CMU-B, etc., and they
used the single letters internally.  Their A machine was the gateway
between the Arpanet and their internal network.  Very quickly, our short
name was removed from the host table, single letter hostnames were
prohibited, and we used "aero" thereafter.

Steve







On Sun, May 3, 2026 at 8:23 PM Noel Chiappa via Internet-history <
internet-history at elists.isoc.org> wrote:

>     > From: Greg Skinner
>
>     > What type of Ethernet (single cable, bridged LAN)?
>
> Even more important, there are two very different 'Ethernets': The
> 'Experimental' 3MB Ethernet, and the 10MB DIX Ethernet. The former
> considerably preceded the latter.
>
>
> Xerox gave a large donation of Altos and associated gear (most notably
> 'Dover' laser printers) to MIT, Stanford and CMU; these all used only the
> 3MB
> Ethernet. (I have been looking online for an original document which says
> when this happened, but I can't find one. Some sources say it happened in
> 1978 , but it might have dribbled over into early 1979.)
>
> Most of Stanford's internal LANs were 3MB Ethernet for quite some time.
> (They
> had their own inter-LAN routers, but I got roped into providing a 'gateway'
> (router) to connect Stanford's internal internet to the ARPANET. I have
> found
> an early version of the config files for it, and it only had a 3MB Ethernet
> interface for quite some time; a 10MB Ethernet was added somewhat later -
> 11/29/1983, to be exact.) I don't know much about what went on at CMU.
>
> MIT only used the 3MB Ethernet for the Altos and the Dover; we had lots of
> our own LANs. Dave Clark and I got into a contest: he was going to write a
> TCP for the Alto, and I was going to produce a router to connect the
> Ethernet
> to the internet in Tech Sq. I have a piece of paper, presented at a
> celebration where Karen Sollins declared the outcome to be a tie, which is
> dated 13 March, 1980. (We must have asked PARC for an Ethernet type for IP
> packets; I see PUP was 01000, and IPv4 was 01001.)
>
>         Noel
> --
> Internet-history mailing list
> Internet-history at elists.isoc.org
> https://elists.isoc.org/mailman/listinfo/internet-history
> -
> Unsubscribe:
> https://app.smartsheet.com/b/form/9b6ef0621638436ab0a9b23cb0668b0b?The%20list%20to%20be%20unsubscribed%20from=Internet-history
>


-- 
Sent by a Verified

sender