Randal Hayes wrote:
> Whenever a communications technology outage is reported and the cause
> is attributed to some problem with a software load or other
> maintenance work performed *in the middle of a workday* I always ask
> myself, "What were they thinking, performing this type of work during
> a busy period?!?"
When the Bell System was trying out prototype ESS gear, they had a
real central office served by their new switch. They had to make some
changes and planned to do so at 3am. However, they hesitated about
taking the switch down at that time, since if a call did come through,
it likely would've been an emergency. Sure enough, a call did come up
and it was indeed an emergency. The engineers realized 24 hr service
had to be truly 24 hour service.
I don't know about today, but ESS was originally built with
two CPUs. One was handling calls and one was reserved in
case of failure or to do maintenance. Early ESS turned out to
be extremely reliable -- the longest outage was on account of a
failed air conditioner. (IIRC this was the Morris IL test).
One of things the big Bell System was able to do was 'beta test' new
hardware and software under very controlled conditions before rolling
it to the whole country. A small community would be selected,
residents notified and trained accordingly, and the system tested.
Not everything passed the test -- certain call features weren't
popular and their initial tone ringers were disliked. The experience
the engineers gained from seeing and maintaining their switch in real
service serving real people was invaluable.