On Thu, 19 Dec 2002, Olin Lathrop wrote:

> > All other things being equal, a cable operated at "97% of
> > capacity" shouldn't fail, so it should be fine.
>
> Of course it ***shouldn't***.  But the point is that good engineers take
> into account what happens when everything isn't exactly as planned,
> especially when there are humans in the loop that may do unpredictable
> things.  The issue is robustness, which is up to you to realize because
> the non-technical customer doesn't know enough to specify it.

I'm in complete agreement with Olin on this one.  My world is filled to
overflowing with things that "shouldn't ever fail", and my job consists
largely of making damn sure we don't take an outage when one or more do
fail, and they do fail regularly.  Or irregularly, which is even worse.
Cable connections, cables, memory, spanning tree protocol, NFS mounts, UPS
power, redundant switches and routers, operating systems, you name it.
All operated well within spec, and all still fail.  Stuff happens.  The
difference between an engineer and a top-notch engineer is often the
ability to design systems that take these things into account, and
balances reliability against cost.

Dale

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics