TR Forums

I didn't check the hours again but I retired that seagate yesterday. It didn't fail, but I ran out of physical space for storage to add, so it was the odd man out.

A 320GB Seagate SATA drive that's been retired into being where Google Drive syncs to. I don't have the # of hours but it's been powered up almost constantly since ~2009.

edit: 66,579 hours.

I have a pair of Seagate 160GB units that still work and have over 80K but they have been retired from active service since 2014. They are used older longitudinal magnetic recording which are much more reliable/tolerant than current PMR/SMR that modern HDDs use.

4x Samsung F1 hard drives out of a Dell T310 server (2010). ~67,000 hours each and no errors. Out of service recently, but impressive hours.

BTW, NVMe (and apparently USB-SATA) don't report accurate power on hours.

Waco wrote:

I have a Bigfoot XT that's well over 130K hours. It was on for a decade and a half or so, and it still powers up just fine. It predates SMART, though, so I have no real proof.

None of the other drives come close, but many of the drives in my NAS are far north of 40K hours (5 years or so).


If I could grab the stats I think the drives in the SGI Indigo^2 from work would win. I retired it a year ago, but it ran 24/7 from 1994 until 2016. That would be...192720 hours of power on time, give or take a few thousand. I could dig up the original purchase order if I was sufficiently motivated, but it was a monster that wouldn't stop working. It took considerable zeal to get the sysadmin to let it go for a Dell R420.

How the hell did a Bigfoot managed not to die on you? The damn things broken down or had issues within 1-3 years of life.

Krogoth wrote:

How the hell did a Bigfoot managed not to die on you? The damn things broken down or had issues within 1-3 years of life.

No idea. It fired up and happily booted Windows 95 last year when we were playing around with the old hardware.

90,899 hours on my 4 disk raid-5 array.

Yep, those 320GB Seagates were built right. 81,188 hours on mine, or 9.26 years of active use. Used as one of four drives in my first NAS. None of them ever failed, sold two and use the others as scratch drives in two rigs to save wear on the SSDs.

I specifically disabled the spindown timer on the NAS, and drive sleep on my OS since just opening any explorer window would always wake drives (networked and local) and I attribute their long life to that.

Ryu Connor wrote:

90,899 hours on my 4 disk raid-5 array.

Jeez, well that beats mine. Did you disable the sleep timer or is that mitigated in other ways for that array?

Waco wrote:

Krogoth wrote:

How the hell did a Bigfoot managed not to die on you? The damn things broken down or had issues within 1-3 years of life.

No idea. It fired up and happily booted Windows 95 last year when we were playing around with the old hardware.

Low RPMs, so less mechanical wear. And they don't build 'em like they used to.

just brew it! wrote:

Waco wrote:

Krogoth wrote:

How the hell did a Bigfoot managed not to die on you? The damn things broken down or had issues within 1-3 years of life.

No idea. It fired up and happily booted Windows 95 last year when we were playing around with the old hardware.

Low RPMs, so less mechanical wear. And they don't build 'em like they used to.

Bigfoots were more prone to issues (headers had a nasty tendency to crash). It is probably a fluke that managed to avoid most of them. LMR HDDs were more reliable in general which is why PMR (more parts and less tolerant to physical stresses) never took off beyond experimental/prototype units back in the 1970-1980s. HDD manufacturers fell back to PMR once superparamagnetic effect start to kick in for LMR units while demand for more data density was stronger then ever. Notice how HDD reliablity and warranties took a noticable drop once PMR was adopt en mass? Not a coincidence. Funny that PMR units are starting to have the same superparamagnetic problem now and HDD manufacturers are looking into even more exotic ways to overcome it (Helium, SMR, HAMR/MAMR) in order to keep up in the data density race.

@The Egg - Very cool! I was thinking of doing something like that myself, but it is currently way, way down on the "todo" list!

bthylafh wrote:

A 320GB Seagate SATA drive that's been retired into being where Google Drive syncs to. I don't have the # of hours but it's been powered up almost constantly since ~2009.

edit: 66,579 hours.

Good thing it wasn';t a 16 bit integer

(i was gonna make a -65,535 joke otherwise)

just brew it! wrote:

@The Egg - Very cool! I was thinking of doing something like that myself, but it is currently way, way down on the "todo" list!

You could use octal or hexadecimal.

pikaporeon wrote:

bthylafh wrote:

A 320GB Seagate SATA drive that's been retired into being where Google Drive syncs to. I don't have the # of hours but it's been powered up almost constantly since ~2009.

edit: 66,579 hours.

Good thing it wasn't a 16 bit integer

(i was gonna make a -65,535 joke otherwise)

It would technically be -32,767.

I think SMART attributes are 48-bit values.

So that's what? Several hundred trillion hours?

Krogoth wrote:

It would technically be -32,767.

-32768

meerkt wrote:

Krogoth wrote:

It would technically be -32,767.

-32768

Given how integers are handled in binary. The last bit is used to determined if the value is negative or positive and how the math works out. You end-up losing a "1" hence when you notate 1111111111111111 it ends converting to -32,767 in decimal for an integer value.

Krogoth wrote:

meerkt wrote:

Krogoth wrote:

It would technically be -32,767.

-32768

Given how integers are handled in binary. The last bit is used to determined if the value is negative or positive and how the math works out. You end-up losing a "1" hence when you notate 1111111111111111 it ends converting to -32,767 in decimal for an integer value.

No, meerkt was right. 1111111111111111b == -32,768 in two's complement notation (which is how all modern systems represent integers)

Nerds.

just brew it! wrote:

Krogoth wrote:

meerkt wrote:

-32768

Given how integers are handled in binary. The last bit is used to determined if the value is negative or positive and how the math works out. You end-up losing a "1" hence when you notate 1111111111111111 it ends converting to -32,767 in decimal for an integer value.

No, meerkt was right. 1111111111111111b == -32,768 in two's complement notation (which is how all modern systems represent integers)

You're right. My mental math just got mixed-up with positive integers in binary (2^x -1).

A 1TB Samsung drive at 12,636 hours. Although It'll be retired later this year for something larger.
I have a long since retired 320GB WD drive that ended it's life at just south of 60,000 hours.

On the SSD side, a 1TB Mushkin drive is sitting at 41,427 total NAND writes.

Rand wrote:

On the SSD side, a 1TB Mushkin drive is sitting at 41,427 total NAND writes.

With external SSDs starting to become much more common (and affordable), there really needs to be alot more discussion/testing on powered-off NAND data retention. Were it not for my memory of the 840EVO debacle, I wouldn't even be aware that it was an issue.

34.833 hours and 41,98TB writes on a Samsung SSD (that 840 line, too). Might have had mechanical drives with more hours but they either crapped out on me, or were replaced for lack of space.

Hours on SSDs don't matter, methinks.

Well, as The Egg mentions, you're almost more concerned about "Power Off" hours

Kougar wrote:

Jeez, well that beats mine. Did you disable the sleep timer or is that mitigated in other ways for that array?

Sleep timer is enabled for the array, though with a long timer (30 minutes).

The load cycle count is 14,006. I don't find that particularly problematic given the age. These are WD Blacks 1TB (1001FALS).

Glorious wrote:

Well, as The Egg mentions, you're almost more concerned about "Power Off" hours

Yep. Still waiting for someone to do a big retention test.
But that's for the "power off hours SSD" thread.

meerkt wrote:

Hours on SSDs don't matter, methinks.

It sort of does if the SSD is the OS drive. It will translate into monthly wear because Windows will SSD defrag those monthly to prevent shadow volume fragmentation issues.

Just one of the factors for why my OS SSD has accrued a fair amount of additional wear compared to my old, small gaming SSDs. Anyone silly enough to use QLC drives will enjoy replacing it in 5-10 years, because hey it's windows NTFS and we can't allow the shadow volume recovery file system to get too fragmented.

The oldest SATA hard drives I have with me:

WD WD20EARS 29,134 hours (0% health)
Hitachi HDT721010SLA360 43,860 hours (100%)
Seagate ST31000528AS 51,596 hours (100%)

I have a bunch more but they're IDE and I don't have anything to check them with anymore.

I have an old PATA 4GB Maxtor drive in a retro gaming rig that came in a computer I bought way back in 1999 with 80k hours on it. All of the other hard drives I own are pretty new aside from one of those 1.5TB Seagate drives that just die. It still worked when I replaced it and frankly I'm afraid to hook it up and power it on. I bought it from some one on TR and used it until about a year ago almost constantly so I'm sure it has at least 50k hours on it. I was one of the chosen few.