Re: OT: The Geek defense

On Thu, 28 Feb 2008 19:35:40 -0000, tim <TimJ@xxxxxxxxxxxx> wrote:

One was the anxiety that accompanied a major production change. On one
occasion I designed a whole lot of changes designed to make the system run
faster, as there were response time problems. On paper they looked great,
and I even wrote and ran a small IMS simulator and it looked good there as
well. However when we stress tested it, it ran fast alright, until IMS
crashed. It lasted about 20 minutes between crashes. It turned out that
the increased intensity of multiprocessing brought out some latent bugs in
IMS and caused the crashes. Eventually IBM fixed the bugs and the changes
went in OK, but I was sweating that weekend.

I remember talking with a Univac SE (our system was a 9030), and he
told me about a program that was crashing. It took a while to
determine the problem - apparently the recent change speeded it up
enough to show up a *hardware* bug.
.