Date   : Thu, 08 Nov 1990 11:44:34 GMT
From   : mcsun!ukc!edcastle!cs.ed.ac.uk!cs.edinburgh.ac.uk!gdmr@uunet.uu.net (George Ross)
Subject: Early microcomputer networks

In article <JIM.90Nov6131135@baroque.Stanford.EDU>, news@helens.Stanford.EDU
(news) writes:
> ...  But what about networks that became extinct, the
> network dinosaurs?  What was happening with other mini/microcomputer
> networks around that time?

They're still alive and kicking (at least, in their later incarnations).

In the early '70s we had an assortment of minis, tied together with point-to-point
"links", running initially at about 256Kbaud, and latterly at 2Mbaud.  These
links were full-duplex flow-controlled, with error detection (parity and
positive/negative acknowledgement).  The machines included several Interdata
70s,
a dozen or so 74s, a PDP-8, a PDP-9 and a PDP-15.  Originally they all
ran their
usual operating systems, but around 1976 we added a big (67Mbyte!) disc
to one of
the 70s and converted it to a file server.  At first it just did file storage
and
transfer, but by the early 1980s we had converted the 74s to be discless
clients
which booted from the central server (thanks to the AUTOLOAD instruction)
and then
relied completely on it and a separate compute server (another 70).  Meantime
the PDP-8 had been retired and the 9 and 15 converted to use the central
server as
a virtual-DECtape server as well as a file server.

Since then we've converted the links to a 2Mbaud CSMA/CD network which
provides reliable virtual circuits in the firmware, added 60-odd 68K-based
machines, retired
all the Interdatas and the PDP-9 and 15, and added several new 68K-based
file servers.
There's nothing left now of the original hardware, though for a while around
the mid
'80s it was all running happily together.  In fact, the last "link" board
was removed
just last year when we switched off our 780.  The file access protocol
is still the
same, though, more or less, as that used in 1976 -- the only difference
is that the current servers don't understand some of the more baroque block-size
constructs needed
to get the virtual DECtapes working sensibly.

It's a stateful protocol, btw, with user validation done by the servers
on the basis
of link- and VC-specific tokens issued by them in response to the initial
logon request.
Just what's needed, in fact, to deal with a labful of students all doing
operating systems practicals.

> 
> Jim Helman
> Department of Applied Physics                     Durand 012
> Stanford University                               FAX: (415) 725-3377
> (jim@KAOS.stanford.edu)                   Work: (415) 723-9127

-- 
George D M Ross, Department of Computer Science, University of Edinburgh
031-650 5147 or 031-667 1081    gdmr@uk.ac.ed.cs (or cs.ed.ac.uk if you prefer)