> I've just seen too many Quantum and Seagate drives fail in large 
> disk arrays to get a warm-fuzzy from them.

What is 'too many'?

If you have lots of drives, lots of them will fail over time.  that's
the way it goes, for all drives.  You can turn an MTBF number into a
'yearly expected failure percentage' number, to figure out how many
you should expect to be losing, even...


> I guess I'm just stubborn :-)

No, you and the others seem to have completely missed the fact that
reliability is largely a model-by-model thing, with modifiers for the
manufacturer, defects in that run, etc.

It's meaningless to say that "drives from manufacturer Foo are
reliable".


I personally have found Quantum drives to be ... rather reliable (I've
owned a 210, 1800 and 2100), though i forget what the exact models of
each are...  On the other hand, there are known 'bad' quantum drives
which i've been lucky to never have, e.g. the infamous 105s (the kinds
shipped in lots of sparcs...  stiction).


Manufacturers publish a MTBF.  There's a reason for that, and, if i
understand correctly, there has to be a minimum number of failed
units before an MTBF can be published, to encourage statistical
accuracy.  As far as i know, MTBF numbers are typically are pretty
accurate.



cgd