Seagate reduces 7200.12 warranty period in the fine print?

[Grant]

Download and compare the manuals for Seagate's 1TB drives, the older
ST31000528AS and the new ST31000524AS. You'll find a new clause in the
product manual (reformat to shorter lines, replace bullet symbol):
"
Document Revision History
Revision Date Description of Change
Rev. A 08/20/2010 Initial release.
Rev. B 03/04/2011 Added Reliability section.
...
2.12 Reliability
The product shall achieve an Annualized Failure Rate (AFR) of
0.34% (MTBF of 0.75 million hours) when operated in an
environment of ambient air temperatures of 25°C. Operation at
temperatures outside the specifications in Section 2.9 may
increase the product AFR (decrease MTBF). AFR and MTBF are
population statistics that are not relevant to individual units.

AFR and MTBF specifications are based on the following
assumptions for desktop personal computer environments:
o 2400 power-on-hours per year.
o 10,000 average motor start/stop cycles per year.
o Operations at nominal voltages.
o Temperatures outside the specifications in Section 2.9 may
reduce the product reliability.
o Normal I/O duty cycle for desktop personal computers.
Operation at excessive I/O duty cycle may degrade product
reliability.

The desktop personal computer environment of power-on-hours,
temperature, and I/O duty cycle affect the product AFR and MTBF.
The AFR and MTBF will be degraded if used in an enterprise
application.
"
So does this mean the newer Seagate 1TB drives no longer only good
for a year in 24/7 operation? Is Seagate going to reject units
returned for warranty once they exceed 7200 hours (3 x 2400) power
on?

The above quote is from the Mar'2011 version of the product manual,
headlining the ST31000524AS drive. The Aug'2010 release of the
7200.12 product manual headlines the ST31000528AS drive.

Note that Seagate drives are good for that 0.34% (1:300?) failures
per years, plus they'll throw an unrecoverable error (lose your data)
each 10^14 bits or 12.5TBytes read. Not looking so good, are they?

Grant.

 

[Arno]

Download and compare the manuals for Seagate's 1TB drives, the older [...]
The desktop personal computer environment of power-on-hours,
temperature, and I/O duty cycle affect the product AFR and MTBF.
The AFR and MTBF will be degraded if used in an enterprise
application.
"
So does this mean the newer Seagate 1TB drives no longer only good
for a year in 24/7 operation? Is Seagate going to reject units
returned for warranty once they exceed 7200 hours (3 x 2400) power
on?

No idea. The 2 year EU warranty will still be valid, I expect.

The above quote is from the Mar'2011 version of the product manual,
headlining the ST31000524AS drive. The Aug'2010 release of the
7200.12 product manual headlines the ST31000528AS drive.

Note that Seagate drives are good for that 0.34% (1:300?) failures
per years, plus they'll throw an unrecoverable error (lose your data)
each 10^14 bits or 12.5TBytes read. Not looking so good, are they?

This is rather optimistic on the drive failure side. Something
around 2%..5%/year is more realistic. On the unrecoverable error
side, I expect things are actually a lot better, like 1 unrecoverable
per 10^15...10^16 bits read.

Arno

 

[Rod Speed]

Grant wrote

Download and compare the manuals for Seagate's 1TB drives, the older
ST31000528AS and the new ST31000524AS. You'll find a new clause in
the product manual (reformat to shorter lines, replace bullet symbol):
"
Document Revision History
Revision Date Description of Change
Rev. A 08/20/2010 Initial release.
Rev. B 03/04/2011 Added Reliability section.
...
2.12 Reliability
The product shall achieve an Annualized Failure Rate (AFR) of
0.34% (MTBF of 0.75 million hours) when operated in an
environment of ambient air temperatures of 25°C. Operation at
temperatures outside the specifications in Section 2.9 may
increase the product AFR (decrease MTBF). AFR and MTBF are
population statistics that are not relevant to individual units.

AFR and MTBF specifications are based on the following
assumptions for desktop personal computer environments:
o 2400 power-on-hours per year.
o 10,000 average motor start/stop cycles per year.
o Operations at nominal voltages.
o Temperatures outside the specifications in Section 2.9 may
reduce the product reliability.
o Normal I/O duty cycle for desktop personal computers.
Operation at excessive I/O duty cycle may degrade product
reliability.

The desktop personal computer environment of power-on-hours,
temperature, and I/O duty cycle affect the product AFR and MTBF.
The AFR and MTBF will be degraded if used in an enterprise application.
"
So does this mean the newer Seagate 1TB drives
no longer only good for a year in 24/7 operation?

Nope, it just means that thats one of the conditions for that AFR and MTBF statistic.

Is Seagate going to reject units returned for warranty
once they exceed 7200 hours (3 x 2400) power on?

They cant do that legally unless that is included in the warranty statement.

The above quote is from the Mar'2011 version of the product manual,
headlining the ST31000524AS drive. The Aug'2010 release of the
7200.12 product manual headlines the ST31000528AS drive.

Note that Seagate drives are good for that 0.34% (1:300?) failures per years,

It says above that that is just plain wrong. Its a statistic, not what an individual drive will see.

plus they'll throw an unrecoverable error (lose your data) each 10^14 bits or 12.5TBytes read.

It doesnt say that either.

Not looking so good, are they?

Depends on what other drives from other manufacturers have for those stats under those condititions.

 

[Grant]

Download and compare the manuals for Seagate's 1TB drives, the older [...]
The desktop personal computer environment of power-on-hours,
temperature, and I/O duty cycle affect the product AFR and MTBF.
The AFR and MTBF will be degraded if used in an enterprise
application.
"
So does this mean the newer Seagate 1TB drives no longer only good
for a year in 24/7 operation? Is Seagate going to reject units
returned for warranty once they exceed 7200 hours (3 x 2400) power
on?

No idea. The 2 year EU warranty will still be valid, I expect.

The above quote is from the Mar'2011 version of the product manual,
headlining the ST31000524AS drive. The Aug'2010 release of the
7200.12 product manual headlines the ST31000528AS drive.

Note that Seagate drives are good for that 0.34% (1:300?) failures
per years, plus they'll throw an unrecoverable error (lose your data)
each 10^14 bits or 12.5TBytes read. Not looking so good, are they?

This is rather optimistic on the drive failure side. Something
around 2%..5%/year is more realistic. On the unrecoverable error
side, I expect things are actually a lot better, like 1 unrecoverable
per 10^15...10^16 bits read.

Their enterprise drives go 10^15 bits, and a 1:137 failure rate for 24/7
operation. Wonderful, at least that's a better (more believable) spec
than MTBF, which is based on failures within population of drives.

 

[Arno]

Download and compare the manuals for Seagate's 1TB drives, the older [...]
The desktop personal computer environment of power-on-hours,
temperature, and I/O duty cycle affect the product AFR and MTBF.
The AFR and MTBF will be degraded if used in an enterprise
application.
"
So does this mean the newer Seagate 1TB drives no longer only good
for a year in 24/7 operation? Is Seagate going to reject units
returned for warranty once they exceed 7200 hours (3 x 2400) power
on?

No idea. The 2 year EU warranty will still be valid, I expect.

The above quote is from the Mar'2011 version of the product manual,
headlining the ST31000524AS drive. The Aug'2010 release of the
7200.12 product manual headlines the ST31000528AS drive.

Note that Seagate drives are good for that 0.34% (1:300?) failures
per years, plus they'll throw an unrecoverable error (lose your data)
each 10^14 bits or 12.5TBytes read. Not looking so good, are they?

This is rather optimistic on the drive failure side. Something
around 2%..5%/year is more realistic. On the unrecoverable error
side, I expect things are actually a lot better, like 1 unrecoverable
per 10^15...10^16 bits read.

[/QUOTE]

Their enterprise drives go 10^15 bits, and a 1:137 failure rate for 24/7
operation. Wonderful, at least that's a better (more believable) spec
than MTBF, which is based on failures within population of drives.

I suggest you stop reading the vendor publications and start
looking at real-world data.

1. I run long SMART selftests on all my disks every 14 days.
With one in 10^14 bits, I would be getting an unrecoverable
error in 2.5 months (2.5TB total online). Instead I have seen
no unrecoverable error at all in > 2 years. I have had instances
of seemingly crashed firmware, that were fixable with a power-cycle
and no data-loss. And this is only the SMART surface scans,
normal reads are not included. They may contribute significantly
to the read data volume.

2. The drive manufacturer MTBF sounds nice, but first, you do not
get the perfect conditions these tests are run under in practice,
and second, real world observations do not agree.

Typically, a rough approximation of 5%/year failure rate seems
a more realistic real-world number. Of course, if you handle
the disks exceptionally careful, cool them well and put no high
load on them, _and_ are lucky to habe bought disks with no
design or manufacturing problems, you may see lower figures.

But real-world numbers also include that a complete
series may have significantly lower reliability (which you
only know afterwards and HDD manufacturers to not repeat
the MTBF tests for every model). We all know that
there are such series, e.g. from Seagate when they started
manufacturing in China or messed up their firmware, or
the venerable IBM "DeathStars".

Example: I know of a 6 x 8 disk installation (RAID6)
that I helped plan and that used the then-new Seagate
ES drives with 1'000'000h MTBF. 4 of those disks failed
in the first year. That is a 8% failure rate or an
observed MTBF of around 100'000h. This was in professional
racks with good cooling, moderate load and in a 22C server
room, hence no identifiable external influence.

Hence, the 5% number includes design errors, manufacturing
problems, rough handling (within reason), cooling issues
and so on. You are well advised to use it for reliability
considerations, not the all-clean, but unrealistic, MTBF.

For disks running hot, you should also derate the MTBF by
a factor of 2 for every 10C above 25C and add that as
additional failure probability to the 5%. An MTBF of 1'000'000
at 25C is only 120'000h at 55C, giving you something like
10% total annual failure rate.

Of course, all my numbers are very rough approximations. But
for realistic risk management, the 5% / year number is sound,
the 0.9% / year number a 1'000'000h MTBF suggests is not.


Arno