Things just weren't as cool in those
days!
We'd been working day and night to complete
software changes and to finish testing a large message switching
system.
It was a long time ago now but I recall
the hardware comprised four or five rows of seven foot racks,
each rack housed in a stove enamelled grey cabinet with chrome
fittings and there were about a dozen racks in each row.
All this lot represented a mere five
computers, working in a group, sharing common memory. The idea
was that if one or two (or more if you were lucky) computers
failed, service would be largely unaffected.
It was a Government Establishment and
of course there were lots and lots of Civil Servants around.
The system had been delivered and its
hardware commissioned a few years earlier, but as with most complicated
systems, software bugs had kept it from being accepted by the
customer.
In the 60s, computers were a bit different
to those around today.
For instance a 32K store occupied two
seven foot, nineteen inch racks....Nowadays, 32Ks worth of memory
would be little larger than a pinhead.
Besides being very big, the system got
fairly hot too.
In fact we needed a proper cooling system.
This consisted of a large heat-exchanger on the roof, a chiller
cabinet with lots of pipes and pumps (occupying one of the seven
foot racks) and a network of ducts feeding the top and bottom
of every rack.
There was a substantial false floor
about eighteen inches deep in which ran the cooling trunking
and masses of cables.
There was also a false ceiling carrying
more cooling trunking and loads more cables.
Power supplies had much the same sort
of voltages as are supplied inside your modern PC but we were
supplying hundreds of amps via large copper bus-bars feeding
all the racks, and within the racks, feeding the various electronic
units.
I think there were some four or five
units across the width of a shelf and probably about a dozen
shelves in each rack. Each unit carried about twenty or so printed
circuit boards and each circuit board had a few dozen transistors,
dozens of diodes and lots of resistors and capacitors. I don't
remember any integrated circuits but there may have been some
in one or two of the newer units. These would have been nothing
more complicated than simple gates or "D type" latches.
The day finally arrived when we were
ready to demonstrate our handiwork.
Tests were scheduled to last for a week.
Later on, most of us had to go away
in case we saw something we shouldn't, such was the nature of
the messages that were being "switched".
The security of the free world hinged
on the weeks results!
Lots of Senior Civil Servants appeared
and gathered round a young Welshman who was the Project Leader.
Our MD arrived with his Project Manger.
It was a very very important day.
For starters we hadn't been paid yet.
If we passed the test we got paid. If we failed there would be
a lot of bother and we wouldn't get paid.
Zero hour was ten o'clock.
It was now 9:45.
The Government Project Manager arrived
and the chief civil servant, Cliff, in charge of the computer
room shook his hand and invited him and his entourage to view
THE SYSTEM.
Cliff pauses by the grey cabinet at
the end of the first row.
Still talking, he half turns and grasps
the chrome handle of the cabinet and turns it.
There was a whooshing sound from the
door seal and they all gaze into the depths of the cabinet.
By pure chance he's chosen the chiller
cabinet and what would you expect in a chiller cabinet?
There before them, sitting on the metal
pipes, which were white with frost, was a row of half a dozen
milk bottles, some half full, others unopened, several plastic
lunch boxes and sundry other items such as half-eaten loaves
of bread and a bag of cream buns.
Cliff turned to look at the centre of
interest and stopped talking.
There was a sudden silence.
I suppose, being a SENIOR civil servant
someone had always brought him his tea and he'd never, in the
three years he'd been there, thought about the logistics of it
all.
He must have been a bit embarrassed
because for a full two minutes he just stood there, with his
lords and masters grouped around him, unable to get a word out,
his face getting redder and redder.
Suddenly the silence was broken by a
crescendo of noise.
A klaxon was sounding in short bursts.
A bell was ringing continuously.
Then without further ado everything
went deathly quiet.
All that one was aware of was a large
blue light flashing on the main power supply panel.
For the first time in three years the
power had gone off!
Everyone knew, except Cliff that is,
that one should never open the chiller cabinet door for more
than a few seconds.
Just long enough to retrieve or replace
one's milk.
The door is sealed, open it for longer
than about 30 seconds and the cooling sensors detect something
is amiss.
Longer still and something is definitely
not right.
Open it for two minutes and the system
shuts down because the sensors believe the cooling system has
failed.
Either a million transistors start to
fry or you remove their power.
The latter was the preferred option
and that's what had happened.
No big deal you might say... Just turn
the power on again and Bob's your uncle, the computers are all
up and running again.
Well in those days it wasn't that easy.
To get the required speed from the old
devices they operated flat out. Their different voltages had
to be applied in a very special way. Lose the negative bias rail
and a transistor will melt its junction.
Although automatic power sequencing
was the order of the day, it wasn't quite perfect.
There were also millions of wire wrap
connections.
Some had reached the stage were they
relied on current passing through them to work... Maybe not many
but enough.
There were diodes in the same shape...
Losing their bias opened their junctions.
Nobody ever turned off a 1960s computer
system of this size without at least a weeks work debugging it
when it came on again.
This one took a week.
It was mainly poor wire wraps and duff
diodes with the odd transistor thrown in.
The security of the free world would
have to wait another week. So would getting paid!