> > > Wouldn't you want the drives to just spin forever? I always understood
> > > that spinning something down and then spinning it back up is what causes
> > > them to degrade/die. The analogy would be that a car that runs it's engine
> > > for 10 years straight never needs major repairs, whereas one that is
> > > turned on, driven, turned off, repeat, will be slowly destroyed.
> Some truth here......but you have determine your usage model.
> Last check a few years back was each power-up or spinup from stop is about
> 10 hours of continious running.
> So in an environment of 8-10-hour business days for 5 days per week.
> Power on plus 10-hour day is about 20 hours.
> 20*5 << 7*24 or it costs you 68-78 hours of extra usage against the MTBF,
> per week or 3/4 of a business week.
> Remember that you have to float the heads and sticking can be an issue for
> stopping. Also with accoustic management or spindle throttles you can
> achieve different results. The assumption above is that we are running at
> max RPM always.
> Best case is to place mount drives (regardless of type) in
> anti-shock/vibration racks on an individual bases. Even spindle sync will
> generate beat-harmonics that moments of inertia frown upon. Remeber that
> 75-100 G is easy to do with impulse.
I guess the heads (do they actually graze over the parking track, or
do the arms lean on something when the disk is speeding up/slowing
down?), the bearings of the disk and the disk engine suffer the most
from spinup/down. What has that got to do with vibration or shocks?
Sometimes a disk in a server runs for ages. If the disk is spun down, it
sometimes never comes up again. What might be a technical reason for
this? I've heard something about 'dust' which collects at the parking
track, and in which the heads will kill themselves.
And if a disk dies at the end of it's lifetime (not disk crash), what is
the part that fails? The engine? The heads? The bearings? The servo? The
Ps, is the parking track equal to track 0, and is the parking track near
the center or at the edge of a disk?
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to email@example.com
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Mon Aug 07 2000 - 21:00:05 EST