Two of my unstressed WD20EARX disks have just crapped themselves almost simultaneously after 21846 hours of operation (or 2.493 years), with just 49 power cycles and disk temperatures never above 45 degrees C. Both disks were manufactured in June 2011, but WD's warranty period for consumer gear is only two years.
The culprit? WD's damn idle3 timer which I wasn't quite aware of until a few days ago; hdparm only says 'power management not supported' and I trusted that to mean 'no spindown'.
That timer contraption parks the disk after just 8 seconds of idleness. Guess what: both of mine shipped with the default setting, 8 seconds, and racked up a load cycle count around 2377000. (That WD disk series is rated for a minimum of 300000 load cycles, certainly not 3 million).
Seagate: see data run! run, data, RUN!
Well, the competition isn't much more reliable. There are four ST2000DM001-9YN164 in my colo'd systems at the other end of the world (climate-controlled datacenter, stable power, no nasties whatsoever).
One of the damn disks is going tilt very, very quickly. After just a measly two
years. See az not happy. Run, Seagate, RUN!
Power_On_Hours 17774 Power_Cycle_Count 24 Power-Off_Retract_Count 17 Load_Cycle_Count 131 Airflow_Temperature_Cel 25 (Min/Max 23/32) Temperature_Celsius 25
And the damning:
Current_Pending_Sector 88 Offline_Uncorrectable 88
And it's deteriorating pretty quickly, about 16 new duds every day or two. Fortunately it's all RAID-1 and the replacement is already ordered.