Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84 exptools; site ihuxn.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!mhuxr!ihnp4!ihuxn!res
From: res@ihuxn.UUCP (Rich Strebendt)
Newsgroups: net.rumor
Subject: Re: apocryphal story about ESS crashes - can someone confirm it?
Message-ID: <1009@ihuxn.UUCP>
Date: Tue, 26-Mar-85 22:40:47 EST
Article-I.D.: ihuxn.1009
Posted: Tue Mar 26 22:40:47 1985
Date-Received: Thu, 28-Mar-85 00:14:24 EST
References: <3957@Shasta.ARPA>
Distribution: net
Organization: AT&T Bell Laboratories
Lines: 101

In response to:

> Many years ago, I heard this story, and I was wondering if anybody
> could confirm it:

The entire story as you have related it is pure bullsh*t.
 
>     Back when this was still an unusual event, a new ESS exchange was
>     installed and turned on.  

A switching office is not "installed and turned on" as a computer is --
the process is called "cut-over" and is done (or was done in the era
described in the posting) with axes and strong men yanking on ropes
at something like 2am.  Perhaps someone who has participated in one 
could relate the process in more detail than I am qualified to.

>     Ma Bell, proud of her new child, invited the
>     press in to be dazzled by modern, reliable, computer technology.

Certainly, a telephone company would be proud to have the latest in
technology (as long as it was cost-effective), but after the consent
decree of 1956 it could NOT be COMPUTER technology.

>     After the obligatory tours and speeches, the photographers wanted to
>     take some pictures.  Someone must have thought that a magtape drive
>     would make a good picture, since it looked like part of a real
>     computer.  

Here is where the BS starts.  The No. 1 ESS was designed and installed
WITHOUT TAPE DRIVES.  The nearest thing to a tape drive was the AMA
(Automatic Message Accounting) machine -- many tracks of PUNCHED PAPER
TAPE.

>     A photographer pushed the shutter, the flash went off, and
>     the ESS promptly crashed.

More BS -- ESS machines do not "crash".   Under failure
conditions they will first "phase" as they attempt to isolate the
failure and reconfigure without losing calls in progress.  If this does
not correct the problem with minimal corrective actions, then more
severe actions are taken progressively.  In the extreme, and very rare,
case that the problem is not recovered from in this way, then Emergency
Action (EA) is entered.  During EA the machine assumes that it is
insane.  Piece by piece it tries to find a working configuration --
sanity checking each configuration until it finds one whose sanity can
be proven, then it returns itself to service.  From the onset of EA
until operation is restored the machine's outage time is measured in
seconds.  The requirement on the design of the machine was that its
TOTAL down time (out of service) during its design lifetime (40 years)
was not to exceed 2 hours.  I repeat -- ESS machines do not "crash".

>     Why did the ESS crash when the flash went off, you wonder?  So did Ma.
>     It turns out that the tape drives had optical sensors for the foil
>     end-of-tape markers.  

Guess what -- most tape drives STILL use optical sensors to see the
reflection of light off of the foil "tape marks".  Indeed, I can 
think of no tape drive I have worked with recently that used any
OTHER way to detect the beginning or the end of tape.

>     In spite of the vaunted reprogrammability of the ESS, it turned out to
>     be too dangerous to fix the software bug immediately (it might break
>     something else).  So, all the tape drives in all the ESS exchanges were
>     fitted with little opaque hoods over their end-of-tape sensors.

Double bull roar.

The idea of "little opaque hoods" over the BOT/EOT sensors on ANY tape
drives is ludicrous.  As to the "vaunted reprogrammability of the ESS",
while the program storage medium for the No. 1 ESS was different from
core or semiconductor memory (magnetic twistor memory cards), it was as
reprogrammable as modern EPROMs.  However, it is true that Bell Labs
has always subjected program patches and new features to extensive
testing (though bugs do slip through the finest of testing nets), so
that a fix for this kind of problem would not be installed
"immediately" -- indeed, the conditions needed to cause this problem to
show up and the low frequency of occurrance of those conditions would
probably lead such a fix to be of very low priority and might make it
into the next generic release if someone had no more important problems
to fix.

> It's a good story, even if it isn't true, but did this really happen?
> If it did, do I have the details right?

Obviously, you do NOT have the details (or even the grossest
information) correct.

Lest I appear a know-it-all spoilsport, I have also heard a similar
story, but (more reasonably to my mind) it was supposed to have
happened in the Comp Center of an insurance company which wanted a
picture of the new computer for the Shareholders' Annual Report to show
what a modern company they were.  The flash was set off during a seven
tape merge sort causing the tapes to simultaneously go into rewind and
abruptly terminate several hours of work.  The story was newsworthy
because the operator tried to kill the photographer.  (Justifiable
homocide?)

Now to await the next cheap shot at AT&T.

					Rich Strebendt
					...!ihnp4!ihuxn!res