steve at sohara.org
Tue Apr 19 19:27:52 UTC 2016
On Tue, 19 Apr 2016 11:52:20 -0500 (CDT)
"Valeri Galtsev" <galtsev at kicp.uchicago.edu> wrote:
> On Tue, April 19, 2016 11:16 am, Lowell Gilbert wrote:
> > "Valeri Galtsev" <galtsev at kicp.uchicago.edu> writes:
> >> Somebody with better knowledge of probability theory will correct me if
> >> I'm wrong some place.
> > Well, you are assuming that the probabilities of two drives failing
> > are entirely independent of each other. The person to whom you are
> > responding asserted that this is not the case. Neither of you
> > presented any evidence directly to that point.
> Correct, we didn't hear proof of one or another. I, however, can not think
> of any physical mechanism that can be involved which will lead in case of
> failure of one drive to failure of another. That is why I assume events
> are (pretty much) independent.
The increased load caused by populating the replacement drive is
one such physical mechanism. Probability of failure in a period of time is
affected by how busy the drive has been in that period.
Next Look at the curves for failure probability against age (or
total activity) and note that the classic bathtub shape and consider what
happens when your whole array is on the steep bit at the end and the first
drive failure happens.
Yes random drive failures well within the expected life are pretty
much independent but the coupling gets stronger as the drives get older due
to the increasing tendency of load induced failures.
Steve O'Hara-Smith <steve at sohara.org>
More information about the freebsd-questions