24 Mar
2010
24 Mar
'10
1:42 p.m.
On Wed, Mar 24, 2010 at 06:21:41PM -0000, arcarlini at iee.org wrote:
Alexander Schreiber [als at thangorodrim.de] wrote:
Yes, that is precisely how I understand the MTBF - if 2 out of 4 drives die
in the first month, well, too bad, that's why it is called _mean_ TBF.
The fact that the manufacturer is willing to give said 5 years of warranty
implies that only a small (but definitely non-zero) number of those disks
_should_ die within the first 5 years.
Most electronics these days have a pretty large
MTBF -
usually, by the time you get anywhere near a significant
fraction of the MTBF the equipment is long obsolete anyway.
For instance, one of the hard drives in my machines has a
MTBF of 1.2 _million_ hours, that's about 136 years. Yes, I
expect it to actually fail a bit sooner (it is a MTBF and not
a guarantee, after all), but I don't expect the drive to be
in service in 5 years due to being replaced by a larger one.
As it happens, the manufacturer actually gives 5 years of
warranty for that one ...
MTBF doesn't mean what you think it means. It means if you have
a bank of (say) 1000 such drives and **you swap them for new
ones every time their pass their service life (5 years in your case)
_then_ you'll experience MTBF/num-drives failures every year, on
average**. So if you don't swap your drives out every time
they go out of warranty,
the MTBF you'll experience is not the MTBF they've quoted.
Clearly an MTBF of that long should mean that it's unlikely that you'll
experience a dead drive in the first five years that you own it.
So just two dead ones in the office last week ... I must be lucky :-)
Well, just because statistics says that on average you get hit by lightning
every 42 years[0] doesn't mean you can't get hit twice on the same day ;-)
Regards,
Alex.
[0] number randomly choosen
--
"Opportunity is missed by most people because it is dressed in overalls and
looks like work." -- Thomas A. Edison