[ih] Historical fiction

[Dave Crocker]

  Dave Crocker
  Brandenburg InternetWorking
  bbiw.net

On 5/12/2012 5:56 AM, Craig Partridge wrote:
> In most case I learned of, the relationship was one of cheerful competion
> between the centers on the same campus, except when every few years the
> main computing guys would attempt a takeover and get beaten back.

Well, since you sorta asked for this...

The Campus /91 was tailored for mass processing.  It even had an area 
with a card reader for submitting jobs directly and a slot that printout 
came through.  The /91 was designed to take up to 2MB of main memory. 
The Campus machine had another 2MB attached through peripheral i/o. 
(Side effect was that the cost of a job varied depending upon which 
memory it ran on, since the latter incurred peripheral i/o charges...)

We, on the other hand, had a home-built timesharing service, called 
TORTOS, with the obvious logo.

One of our users that I was friendly with said she took jobs to the 
Campus machine that were simple to run, because the Campus machine had 
lower rates.  But she brought the special handling ones to us.

I didn't know it at the time, but operator training was quite different. 
  Our facility trained everyone to the level of shift supervisor.  (For 
us part-timers, some weekend shifts had us working alone.)

When I switched to the Arpanet project in computer science, my first 
assignment was to help with the planning for the Arpanet's coming out 
party in September, 1972 (Washington DC Hilton) and that meant lining up 
demo programs.  This included a meeting at the Campus facility one 
Saturday and several of us sat around a table in the machine room. I 
think Vint ran the meeting.

At one point there was some commotion over by the operator console.  It 
wound up going on for some 10s of minutes, with phone calls and 
consultation.  I finally asked one of the Campus folk at our table what 
was going on and he said the machine had crashed and they were talking 
to the facility manager about what the problem probably was and whether 
it would need a service call.

Without any thought to etiquette I said they had a bad memory module 
that needed to be replaced.  It was obvious from the pattern of red 
(hardware) lights on the top of the console.  Real hardware errors would 
light only a very few bulbs but memory problems would trigger a random 
pattern of many.

The guy looked at me rather strangely.  I don't remember whether he said 
anything to the folk over at the console.

That was the first time I realized what it meant to train for 
streamlined operations and costs, versus how we had been trained over at 
the Health Sciences machine...

d/