|
|
Posted by peterh5322 on 07/14/06 06:12
On 2006-07-13 19:05:11 -0700, "Toby" <kymarto123@ybb.ne.jpp> said:
> MTBF don't mean shit when the F happens to you.
I was formerly employed by a large mainfame computer manufacturer.
We also made disk controllers and storage subsystems.
We bought our drives from a related division. Because these were
"count, key, data" drives, and not "fixed block architecture" drives,
these were rather specialized items.
That division always provided a so-called "test report" with every drive.
The drives had a specification of something like 5,000,000 hours.
In one shipment, this being numerous pallets full of drives, EVERY
drive had a signed and dated test report which stated each drive passed
all tests, and yet each and every drive was bad.
We later discovered that:
1) the MTBF was calculated using "fuzzy math", and was not supported by
anything even remotely scientific; the real MTBF was way under
1,000,000 hours, and
2) they had an internal directive which, broadly interpreted, meant
"Ship Shit, Don't Slip Ship". which meant it was entirely possible,
even entirely likely, that every drive they shipped could be defective,
irrespective of the "test report"; obviously, those test reports were
faked.
Although we were ahead of our chief competitor at this point, delays in
our drive "pipeline", in order to cure the two not insignificant
problems enumerated above, resulted in us losing any "time to market"
advantage, and we never made any money on that product.
The division was eventually shuttered.
Navigation:
[Reply to this message]
|