[funsec] Database design.

Blue Boar BlueBoar at thievco.com
Tue Jun 6 18:07:06 CDT 2006


Drsolly wrote:
>> And you think that's the MTBF you'll get under your use conditions, huh?
> 
> Yes.

Really?  OK, so here's the possibilities I see.  Please pick one or tell 
me what I missed:

-You know precisely what Maxtor's testing conditions are, and those 
happen to exactly match the conditions under which you use your drive.
-Maxtor is actually being conservative, and their 1.2M hr claim reflects 
typical field usage.  Marketing claims be damned, they're going to tell 
the truth.
-You've done your own study on the 1.2M hr drives, and came up with the 
same numbers

>> (Are you're sure that's not the MTBF for drives that aren't powered on?)
> 
> If you don't power a drive on, it never fails, so the mtbf is infinite.

Oh, I don't know.  I think I could look at a drive that had never been 
powered on, but rusted through after the first 50 years, and declare it 
failed.

See, when they publish a MTBF number, I tend to assume that they did 
things like tested with a constant temperature in a vibration-free 
environment, extrapolated beyond what they should have with the sample 
size and length of test, and then failed to publish the margin of error.

But then, I've been accused of being a pessimist.

						BB


More information about the funsec mailing list