verticalrate wrote:
You may be a software developer but you apparently don't have any experience
in embedded high reliability systems. High reliability is not measured in
Mean Time Between Windows Reboot. High development costs due to a rigorous
development process plus small market size equals high prices per unit.

Another thing to mention is that he may use a small number of computers
for his development work. When you are looking after hundreds of
computers, you can gather rough seat-of-the-pants reliability statistics
very quickly.

I now actually believe-in-my-gut thar hard drive failures *are* discrete
random events with their bell-curve (bi-nomial?) distribution centered
roughly around the MTBF that is published in the manual. Discounting
DOA drives, I've had people get unlucky with HDDs that fail after mere
months of operation, and I've had drives that were made 10 years ago
that keep writing, storing, and reading sectors without trouble.
There's no way that I can predict the failures.

-Luke

P.S. The above is great material for the back-up-your-stuff-please
speech! :-)