Archived from groups: comp.sys.ibm.pc.hardware.storage (
More info?)
Previously larry moe 'n curly <larrymoencurly@my-deja.com> wrote:
> chrisv wrote:
>> What's the scoop on these? They cost a bit more than regular WD
>> drives, but have the fluid bearings and claims of higher reliability.
> I'd like to know why MTBF ratings keep going up, now over a million
> hours for some drives, but the expected lifespan remains at five years
> (and that may be for only eight hours of use per day). Because if
> drives really are better now, shouldn't their expected lifespans also
> be higher?
No. The "component life" is the time the MTBF is valid. The MTBF
is not a lifetime bat a failure probability:
1 / <MTBF> = <propapility of failure per hour of operation>
It is valid as long as the device is younger than the "lifetime".
After that the device enters a "wear out" period where the
failure rate per hour increases. Actually a new device also
has higher failure rate, ofteh called "infant mortality", but
this is usually not shown today. It can usually be avoided by
doing a "burn in", e.g. a week of running hot (accelerates ageing)
under full load.
Since the components of the drive age, the MTBF eventually rises.
Complex epelctronics usuelly have 5 years lifetime, mostly because of
electrolyte capacitors that do not live too long but are far cheaper
than, e.g., long-loved ceramics. Semicondictors run cold (e.g. 40C)
have something like 30 years lifetime. It halves every 10C or so,
regardless of whether the device is operating or not. The base
temperature may be significantly different for power semiconductors
(also current PC CPUs). For mechanics it depends. They are usually
manufactured just well enough that they do not bring the system
lifetime down.
> I thought that all new WD drives now use fluid bearings. I know that
> the 120GB one I bought this Labor Day does.
The bearings are not the only thing that fails.
Arno
Facebook
Twitter
Instagram