Mean time before failure claims for hard drives

[Daniel Prince]

Hard drive manufacturers used to claim ridiculous MTBF (Mean time
before failure) claims. Some of them would claim as much as 500,000
hours. That would be over 57 years. Do hard drive makers still
claim very high MTBF figures?

It seems to me that the hard drives we can buy today have shorter
lives than the drives we could buy ten years ago. I would be
surprised if the average life span of today's hard drives was more
than five years.

 

[VanguardLH]

Hard drive manufacturers used to claim ridiculous MTBF (Mean time
before failure) claims. Some of them would claim as much as 500,000
hours. That would be over 57 years. Do hard drive makers still
claim very high MTBF figures?

It seems to me that the hard drives we can buy today have shorter
lives than the drives we could buy ten years ago. I would be
surprised if the average life span of today's hard drives was more
than five years.

I haven't seen a claim a MTBF of 500,000. I have seen 50,000 and 5.7
years which sounds about right. Some HDD makers have bugs in their head
load/unload count. I've seen reports by Samsung HDD users that this
count increments ridiculously rapidly due to a defect in their handling
of power management under APM. In fact, I don't see HDD makers showing
a MTBF anymore but instead just specifying their warranty period.

Claims of 300,000 to 600,000 are for load/unload cycles (of the heads
on/off the media). Heads can be unloaded without powering off. Have
you looked at the Power-Off (or Emergency-Off) Retract Count (not the
Power-Off Count) in the SMART data for your HDD? Your HDD may not
report this data (attribute #188).

Do you use a power scheme that spins down/up your HDD(s) (i.e., puts the
HDD in low-power or sleep mode)? For a desktop, do you really need to
spin down your HDDs? Laptops are configured to do it to save on power
and increase battery life. You don't have batteries in your desktop so
do you really need to save a nickel per month instead of leaving your
HDDs always spinning? The more often the HDD sleeps, the faster the
retract count increments.

Do you leave your hard disk constantly powered up and do you use a huge
UPS to ensure the device remains powered up all the way through all
power outages no matter how long and how's the regulation from your PSU
and is the device running in a temperature controlled environ? Or,
conversely, you never power down your computer and you have a big UPS to
ensure it never powers down and you have a high-quality low-ripple PSU
and you keep the device in a computer lab that is temperature and
humidity controlled? Or are you like most consumer-grade consumers
where you power down your computer if you'll be away from it for one or
several days or you don't have a UPS that won't keep the computer up if
the power outage is over 4 hours long or your only temperature control
is the wall thermometer controlling your furnace and A/C which means
there is hystersis in temperature control?

 

[Rod Speed]

Hard drive manufacturers used to claim ridiculous MTBF
(Mean time before failure) claims. Some of them would
claim as much as 500,000 hours. That would be over 57
years. Do hard drive makers still claim very high MTBF figures?
Yep.

It seems to me that the hard drives we can buy today have
shorter lives than the drives we could buy ten years ago.

Don't agree with that at all.

I would be surprised if the average life span
of today's hard drives was more than five years.

Havent had any die personally in decades now and
I have a hell of a lot more than I used to have too.

OTOH I have just had one fail in a mate's Toshiba
laptop, literally just 2 weeks inside the warranty.

 

[Franc Zabkar]

Hard drive manufacturers used to claim ridiculous MTBF (Mean time
before failure) claims. Some of them would claim as much as 500,000
hours. That would be over 57 years. Do hard drive makers still
claim very high MTBF figures?

MTBF (Mean Time Between Failure) has a counterintuitive meaning.

AIUI, an MTBF of 500,000 hours means that in a population of 500
drives, say, it is expected that there will be 1 failure every 1000
hours.

- Franc Zabkar

 

[Arno]

Hard drive manufacturers used to claim ridiculous MTBF (Mean time
before failure) claims. Some of them would claim as much as 500,000
hours. That would be over 57 years. Do hard drive makers still
claim very high MTBF figures?

That is not a problem. The MTBF is the inverse failure chance per
time, valid only during the component lifetime. That lifetime
is typically 5 years, but may be shorter.

It seems to me that the hard drives we can buy today have shorter
lives than the drives we could buy ten years ago. I would be
surprised if the average life span of today's hard drives was more
than five years.

When you use the wrong interpretation of numbers, you
will arrive at wrong aparent meanings.

Arno

 

[Arno]

MTBF (Mean Time Between Failure) has a counterintuitive meaning.

AIUI, an MTBF of 500,000 hours means that in a population of 500
drives, say, it is expected that there will be 1 failure every 1000
hours.

Yes. And it is only valid during the component lifetime.
If, for example, you have 3 years component lifetime
and of yout 500 drives, all die after 3.1 years, but
none before, you get an MTBF of infinity.

Also note that the MTBF is often not for extreme
conditions, but for best-case ones.

Arno