Hard Drive Lifespan: Seatools for DOS

[bigbeark]

I recently bought a box of 20 plus hard drives, a mixture of IDE and SATA.

I downloaded all the Drive Tools I could find from Seagate, Western Digital, Samsung, Hitachi and Maxtor.

The best tool by far was Seagate's Seatools for DOS. It evaluates any manufacturers drives, and allows error correction on Seagate (and Maxtors made by Seagate) drives.

Since WD and Samsung only evaluate defects, which Seatools does already, why bother?

At any rate the POH (powered on hours) was very interesting to me. Most of the drives had POH hours between 5000 and 18000. Some of the drives with unfixable media faults had hours as low as 1800. There were several newish drives with I guess bad logic boards as the drive was not detected at all. There were three or four newish drives with the "click of death", the heads refusing to park properly.

Fortunately there was a 750GB with less than 1000 hours, two 250GB drives with no faults, one Seagate 250GB with one error that I corrected, a working 160GB laptop SATA drive, and surprisingly a couple of those skinny Maxtor IDEs which had big hours but no faults.

What's your longevity record for consumer grade IDE and SATA drives.

I still trust SCSI, median average failure time 1.5 million hours!

Since there are only 8760 hours in a year isn't that 171 years?

[lti]

Re: Hard Drive Lifespan: Seatools for DOS

I have two old Seagate drives with no bad sectors. One was made in 1998 and has a high Ultra ATA CRC error rate because the computer's FSB was set to a higher frequency than the chipset could handle. The other drive makes strange metallic ringing noises.

I also have an IBM 20GB laptop drive that was killed by a cheap USB enclosure. Windows only sees a single empty folder, but 1.5GB of space is being used. I also have a Toshiba drive that developed bad sectors after almost exactly two years.

[severach]

Re: Hard Drive Lifespan: Seatools for DOS

There's no reason the software can't run on other brand drives other than to annoy the user. Error correction is not magic performed by the software. The firmware on the drive is where the magic happens and no special software is required to make it work. Just write a sector and the firmware does the rest. All the software does is to read the whole disk and write zeros to any bad sector it finds. This is a good technique if you are trying to recover data and there's a bad sector interfering. It's a bad technique if you're trying to validate a drive as fully functional.

Writing zeros takes the same amount of time as reading the drive but is much more comprehensive. The drive will fix and remap all bad sectors including those that are in places the operating system never uses and at the end the drive sorts the bad sector list so the drive works faster. If there are too many errors then the operation will halt and you know the disk is bad. The tools trying to fix one sector at a time can make the bad sector map so convoluted that the drive will give up. Zeroing often fixes this problem.

The drive will fix bad sectors on any write but only zeroing improves the speed. The drive will not rearrange anything when mersenne twister, random, imaging or any other random data is written.

[PCBONEZ]

Re: Hard Drive Lifespan: Seatools for DOS

I still trust SCSI, median average failure time 1.5 million hours!

Since there are only 8760 hours in a year isn't that 171 years?

That's not what it means.
It's not an estimate of longevity.
http://www.faqs.org/faqs/arch-storag...ction-151.html
.
.
To try and clarify.
Lets say the bearings are good for 5 years 24/7 so they make 5 years the expected lifetime.
The MTBF basically tells you the odds of it failing - before the 5 years is up. [Before the wear-out phase.]
Once you hit the 5 years it's expected to die soon regardless of anything else.
.
A 1.5 million hour MTBF means if you have 1.5 million drives under 5 years old you can expect to loose, on average, one per hour.
And if you have 0.5 million drives under 5 years old you can expect to loose one every 3 hours, on average.
.

[EagleTG]

Re: Hard Drive Lifespan: Seatools for DOS

Your study on a bit larger scale:

Interesting read, if a bit long. There are some good summaries of the study out there too, like http://www.engadget.com/2007/02/18/m...resting-thing/

Sadly Google took the high road and opted to not call drive manufacturers out by name (even though they saw a strong correlation of failure to manufacturer). I'd love to see if any of my suspicions from personal experiences with different manufacturers would be confirmed.

[dood]

Re: Hard Drive Lifespan: Seatools for DOS

I don't bother with any of the other manufacturers tools - I just use SeaTools. In Windows, I use HDTune Pro.

[Pentium4]

Re: Hard Drive Lifespan: Seatools for DOS

I was reading about this on Seagate's website...where in the program does it tell you how many hours are on the drive?

[bigbeark]

Re: Hard Drive Lifespan: Seatools for DOS

The online windows version does not give you the POH (powered-on hours) or let you fix defects.

You have to download the Seatools for DOS program. burn it to CD/DVD/Floppy, then reboot your computer. The program runs in Ram.

[Pentium4]

Re: Hard Drive Lifespan: Seatools for DOS

The online windows version does not give you the POH (powered-on hours) or let you fix defects.

You have to download the Seatools for DOS program. burn it to CD/DVD/Floppy, then reboot your computer. The program runs in Ram.

Yeah that's the one I got. This HDD has 35,000 hours!!!!!!! And only made in February, 2007

[dood]

Re: Hard Drive Lifespan: Seatools for DOS

35,000 hours is almost exactly 4 years. So, have you been running the drive 24/7 since you got it?

[PlainBill]

Re: Hard Drive Lifespan: Seatools for DOS

<SNIP>

I still trust SCSI, median average failure time 1.5 million hours!

Since there are only 8760 hours in a year isn't that 171 years?

You are confusing interface with quality of construction. At heart, a hard drive is an electro-mechanical assembly with an interface to the outside world. At one time SCSI was the favored interface for high performance drives. If I recall conversations with my sons correctly, that is no longer the case, fiber is the preferred interface. If you take an exquisitely engineered and constructed set of platters, bearings and heads and slap an IDE interface it will still be a reliable drive.

If you take something that was a rush design job, slapped together by a bunch of careless slobs using substandard materials it doesn't matter which interface the drive has, it's going to die early and ugly.

PlainBill

[pfrcom]

Re: Hard Drive Lifespan: Seatools for DOS

If you take an exquisitely engineered and constructed set of platters, bearings and heads and slap an IDE interface it will still be a reliable drive.

But such a drive was unlikely to be offered as IDE, which needed to be built to a price - was only likely to show up as SCSI, where its asking price could support the quality of construction

[smason]

Re: Hard Drive Lifespan: Seatools for DOS

You are confusing interface with quality of construction. At heart, a hard drive is an electro-mechanical assembly with an interface to the outside world. At one time SCSI was the favored interface for high performance drives. If I recall conversations with my sons correctly, that is no longer the case, fiber is the preferred interface.

While I agree in theory, in practice I've seen next to NO failures in SCSI and SAS (serial attached SCSI) drives in servers, while I've seen tons of failures of SATA and IDE drives. SCSI and SAS just seem to be better built, whether there's any truth to it, I dunno.

Haven't seen fibre used in SANs lately. SAS drivrs seem to be the mainstream, and iSCSI over Gigabit and 10Gig copper is the latest greatest for SANs as the switches are way less expensive than the fibre ones.

[momaka]

Re: Hard Drive Lifespan: Seatools for DOS

At any rate the POH (powered on hours) was very interesting to me. Most of the drives had POH hours between 5000 and 18000.

Beware that some drives don't report their Power-On Hours correctly. In my experience, Maxtor HDDs are the biggest offenders. Some newer Hitachi and Samsung drives also mis-report the POH - they usually show much higher than normal.
On the other hand, old IBM/Hitachi are very accurate. I've timed both my 305030 and 307020 IBM Deskstars and they count the hours very accurately. The 305030 just turned 23078 hours today. It has 20 reallocated sectors. I found it in a computer about 2 years ago. Back then it had 22028 hours and 19 bad sectors. Not too bad for an old Deskstar .

[PCBONEZ]

Re: Hard Drive Lifespan: Seatools for DOS

But such a drive was unlikely to be offered as IDE, which needed to be built to a price - was only likely to show up as SCSI, where its asking price could support the quality of construction

That's not why.
The reason is there is no point in building an IDE drive faster than 7200 RPM.
The higher RPMs is what necessitates the higher build quality.
.
His point was that -IF- they used the same mechanicals the interface would be irrelevant to failure rate.
.

[shovenose]

Re: Hard Drive Lifespan: Seatools for DOS

That's not why.
The reason is there is no point in building an IDE drive faster than 7200 RPM.
The higher RPMs is what necessitates the higher build quality.
.
His point was that -IF- they used the same mechanicals the interface would be irrelevant to failure rate.
.

Maybe you can help out here PCBONEZ. I don't understand how faster spin speeds of a hard drive correlates with reliability!

In fact, personally I've observed quite an interesting thing regarding those ancient ~20GB IDE 4200RPM 2.5" drives made by Fujitsu, Hitachi, IBM, etc: they last much longer than modern hard drives. They technically aren't more reliable (because their lifetimes and failure method) vary widely but they do last longer...(compared to modern 5400/7200RPM 2.5" SATA/IDE laptop drivers).

Can you explain that?

[PCBONEZ]

Re: Hard Drive Lifespan: Seatools for DOS

Faster is more stressful on the mechanicals so for faster:
- They use higher quality parts/materials. [Better alloys and platings.]
- The tolerances are tighter for the head to hit the 'right spot' on the disc at a higher RPM so:
- - - The machining has to be more accurate.
- - - Faster wear from more rotations/time lessens time to out of spec tolerances.
- - - The firmware also has to be more accurate.
- - - As to cost, there is a lower yield on parts coming out of the 'machine shop' and electroplating phases of production due to the tighter tolerances.
~~~
So higher RPM drives are built to higher quality standards.
[And that's what makes them more expensive.]

----
A secondary thing for older and smaller drives it that they use[d] the Longitudinal Recording Method which is less prone to wear related failures than the newer PRM [Perpendicular Recording Method].
- With Longitudinal Recording the 'spots' on the disc are -physically- larger.
- With PRM the 'spots' are smaller -and- the 'free' space between the 'spots' is also smaller.
-
With PRM the head placement has to be more accurate and that inherently means wear issues are more likely cause problems.
-
With PRM the charge [on the spot] is also somewhat weaker due to less space to put the charge in -and- as they are also closer together -and- as there is more adjacent area spot-to-spot: they are more susceptible to corruption due to charges 'discharging' or 'neutralizing' each other through the free space.
[This is basically the same magnetic media problem tape media is known for. The charges eventually dissipate with time.]
Not a huge problem short term but if you are in the habit of using drives as deep storage back-up for files that don't get rewritten for ages [overwriting refreshes the charges] then a PRM drive isn't the best choice.

---
You also might note that very high RPM drives tend to be smaller capacity.
That's not a marketing thing...
It's because it's easier to keep a smaller 'spinning mass' within dynamic tolerances.
The discs tend to be physically smaller for less spinning weight for the sake of stability while spinning.
.

[Pentium4]

Re: Hard Drive Lifespan: Seatools for DOS

35,000 hours is almost exactly 4 years. So, have you been running the drive 24/7 since you got it?

Yeah pretty much, it's my neighbors computer. The hard drive is dead. Not completely dead, it is recognized by the computer and everything but there are too many sector errors on it for any program to fix. It found 100 errors 55% through the scan and can't go any further. It's had a good life though

[Pentium4]

Re: Hard Drive Lifespan: Seatools for DOS

Okay so I just had to share this with everyone. I'm checking all the hours on my IDE hard drives, the old Pentium 3 comp my neighbors gave me had an old WD hard drive. Made in June, 2000 it has 63,713 hours on it! That's crazy. And they stopped using it in 2008 so it was on almost every minute from 2000-2008

[Agent24]

Re: Hard Drive Lifespan: Seatools for DOS

Personally I don't use Seatools or any manufacturer-specific tool unless it's for a specific reason.

The reason is that Seatools doesn't give a damn about SMART, unless it's 'officially' tripped.

By that I mean that if your hard drive has 500 reallocated sectors, yet 700 must get reallocated for SMART to tell you the drive is faulty, Seatools will tell you there is nothing wrong - yet with that many reallocated sectors there most certainly is!


For this (and other reasons) I use HDAT2 for testing\scanning HDDs. It gives a readout of the full SMART data (Seatools doesn't) and uses common sense to determine if something is bad, rather than just blindly believing the drive.

For a quick SMART readout under Windows, I like Speedfan. Bonus is it can read all your temperatures too, so I have it running anyway.