How many power on hours do hard drives last?

[millusions]

hello!

not exactly Kubuntu question, but it's all happening on kubuntu

i've been playing around with Smart utilities on my server, it has four hard disks, 2 of them are Samsung 2TB each, what ive noticed is that one of them has:
Power On Hours 13147 , the other has just over 8000, the other drives a lot less.
so 13147 divided by 24 hours is about 547, so the drive has been on for 547 days if that data is correct, its quite possible, i dont really pay attention to time unless i have to be somewhere..

anyway, should i be looking to replace those drives?

is this a cause for concern?

[kubicle]

Power on hours is a bad metric to use when predicting harddisk failures. There are too many variables involved to give any accurate predictions. Hard disks commonly last about 3-5 years (depending on whether you're talking about averages or medians), but that still leaves out type of usage (continually running or frequently stopped, moving or stationary, environment etc.) and differences between individual hard drives. One could say that hard drives last between 5 seconds and 10+ years.

My general advice is to not to replace drives until they show actual signs of breakage (I/O errors, weird noises or quickly increasing numbers of bad sectors, for example), unless you have other reasons to replace them (getting bigger ones, or SSDs, for example). Of course you should keep regular backups of all your data you wish to keep, this includes both new and older drives.

[GreyGeek]

There was/is (?) a metric called "Mean Time Between Failure", which is how many hours a group of HDs will run before half of them die. The last I read it was 50,000 hours. IOW, out of 1,000 HDs about half of them will fail within 5 years 8 months. How many hours a day they have to be on, how many times they are turned on and off, the metric doesn't specify. I also don't know how accurate the metric is, or if it was just a sales tool created by the PR departments of the manufacturers.

Wikipedia discusses several metrics and a study by Google in 2007, which has surprising results:

A 2007 study published by Google suggested very little correlation between failure rates and either high temperature or activity level. Indeed, the Google study indicated that "lower temperatures are associated with higher failure rates". Hard drives with S.M.A.R.T.-reported average temperatures below 27 °C (81 °F) had higher failure rates than hard drives with the highest reported average temperature of 50 °C (122 °F), failure rates at least twice as high as the optimum S.M.A.R.T.-reported temperature range of 36 °C (97 °F) to 47 °C (117 °F). The correlation between manufacturer/model and failure rate was relatively strong. Statistics in this matter are kept highly secret by most entities — Google did not relate manufacturers' names with failure rates, though they have since revealed that they use Hitachi Deskstar drives in some of their servers.

Google's 2007 study found, based on a large field sample of drives, that actual annualized failure rates (AFRs) for individual drives ranged from 1.7% for first year drives to over 8.6% for three-year old drives. A similar 2007 study at CMU on enterprise drives showed that measured MTBF was 3–4 times lower than the manufacturer's specification, with an estimated 3% mean AFR over 1–5 years based on replacement logs for a large sample of drives, and that hard drive failures were highly correlated in time.

A 2007 study of latent sector errors (as opposed to the above studies of complete disk failures) showed that 3.45% of 1.5 million disks developed latent sector errors over 32 months (3.15% of nearline disks and 1.46% of enterprise class disks developed at least one latent sector error within twelve months of their ship date), with the annual sector error rate increasing between the first and second years. Enterprise drives showed less sector errors than consumer drives. Background scrubbing was found to be effective in correcting these errors

My own experience is that since my first PC that had an HD (in addition to the floppies) in 1993 1983, I have never had an HD go bad on me. That's probably 10 HDs. I've had a couple come close to using all the reserve sectors though.

[dibl]

Here is a 7- or 8-year old hard drive -- an inexpensive consumer-grade drive, found in mass-produced desktop systems such as the e-Machines box that it is installed in:

root@decembersid:~# smartctl -ia /dev/sda
smartctl 6.2 2013-07-26 r3841 [x86_64-linux-3.16-0.towo-siduction-amd64] (local build)
Copyright (C) 2002-13, Bruce Allen, Christian Franke, www.smartmontools.org

=== START OF INFORMATION SECTION ===
Model Family:     Hitachi Deskstar T7K500
Device Model:     Hitachi HDT725025VLA380
Serial Number:    VFC100R1CUURVM
LU WWN Device Id: 5 000cca 210cbbde4
Firmware Version: V5DOA52A
User Capacity:    250,059,350,016 bytes [250 GB]
Sector Size:      512 bytes logical/physical
Device is:        In smartctl database [for details use: -P show]
ATA Version is:   ATA/ATAPI-7 T13/1532D revision 1                                                                          
Local Time is:    Fri Aug 15 15:53:17 2014 EDT                                                                              
SMART support is: Available - device has SMART capability.                                                                  
SMART support is: Enabled

And here is the power-on line from the test:

  9 Power_On_Hours          0x0012   092   092   000    Old_age   Always       -       56177

Other than being a slow drive by today's standards, it seems fine -- the test results indicate no problems. But I wouldn't trust my important data on it without a frequent backup.

[SteveRiley]

I have never had an HD go bad on me.

It's happened to me once in 25 years. The sound of a head crashing into the platter is something you'll never forget...

[millusions]

Thanks guys! you're absolutely right!

thinking back, its true, most drives develop bad sectors rather than just stop dead,
i do vaguely remember whining noises, as in spinning up repeatedly..

that's a pretty cool wikipedia article
i was wondering why Hitachi drives are just a little bit more expensive from my local Pc shop.


thats a nice 56k dibl

i guess im freaking out as these drives dont have backups, maybe i should think of something, i would hate to lose my 28 seasons of Survivor lol

[Silent Observer]

Another data point -- I've had precisely one hard disk fail in service without obvious "abuse" in the 27 years I've owned hard disks. My first hard disk required regular backup/low level format/restore cycles, but that was a well known "feature" of CMI hard disks in the mid-80s, especially those built for MFM being used with RLL controllers. Since then, I've lost two hard disks that were in a tower that fell out of the back of my van, about four feet drop, onto concrete (that motherboard eventually failed also, in a way I believed was due to cracked traces); I don't count those. The one no-obvious-cause failure I've had was a Seagate 1 TB drive that started popping up a S.M.A.R.T. alert during startup just about two weeks after the one year warranty expired; it showed no signs of temperature problems or other issues that showed in the S.M.A.R.T. reports, just started developing read errors -- and S.M.A.R.T. gave enough warning I was able to replace the drive before any data was lost (I bought the same model as a replacement, largely because that was what was available locally at a good price -- and I have no reason not to trust Seagate). FWIW, the other drive currently in my main machine is almost ten years old, I've got a 40 GB in my second desktop that dates from 2002 or so, and the 4 GB that came with my 1997 laptop is still usable (I replaced it with a "redneck SSD" made from a Compact Flash unit made for a camera when I rehabbed the laptop with Linux).

I've had drives that ran for a couple years with bearing noise; as long as the RPM stays steady, noisy bearings don't seem to cause trouble (other than potentially driving the user into conniptions). If you've got a drive that's giving warnings, though, and it contains something important (not sure I'd class a bunch of Survivor in that category, but tastes differ), I'd point out that I've been seeing 3 TB units under $100 lately, and you could set up a RAID 1 (two drive redundant storage) with 1 TB drives for not much over $100.

[GreyGeek]

There was a period of time, about 15 years ago, during which Western Digital (IIRC) was in financial trouble and began sending out defective, returned HDs to fill new orders. Eventually they were forced to recall 400,000 HDs.

Also, back in that day and before, the way to do low level formats was to open a console, run debug, and at the debug prompt enter G = C800:5. Some manufacturers used different entry point numbers but IBM, WD and a lot of the others used that hex setting. It was a bad WD drive which could not be low level formatted using that command with debug.

[dibl]

There was a period of time, about 15 years ago, during which Western Digital (IIRC) was in financial trouble ...

Yep, I remember that. Although in the past 5 or 6 years I have bought WD drives exclusively, and with no regrets (knocking on wood).

But these things seem to go in cycles. Back in the early- and mid-1990s, Maxtor was a very well respected drive manufacturer. Then their quality control went south and the rest of the company followed soon thereafter and the remains were picked up by Seagate. And Quantum had good drives for some years before they sold the business to Maxtor, and then Miniscribe, etc. etc.

[GreyGeek]

Ah, Maxtor! That was the drive I was trying to remember but couldn't. It was the worst for sending out defective drives to fill new orders.

[MepisReign]

Who remembers Quantum? IMO those were reliable HDDs

MR

[Snowhog]

Not truly scientific, but at least a good write up on 'their' experiences.

Who makes the most reliable hard drives?

[millusions]

Ah, Maxtor! That was the drive I was trying to remember but couldn't. It was the worst for sending out defective drives to fill new orders.

yup, i remember sending Maxtor drives from Australia to Singapore for warranty




Cool article Snowhog

[millusions]

, and it contains something important (not sure I'd class a bunch of Survivor in that category, but tastes differ),.

it's the greatest show available,

you get a bunch of people in some remote location,
you starve them a little,
throw in some physical exertion,
then dangle 1 million dollars in front of them,
film the results
loop

[Silent Observer]

I remember using similar commands (I think it was "jump" in a DOS debugger, though I don't recall the address) to low-level format my old MFM and RLL drives in the '87 to '90 time frame. I wasn't aware the option had still existed as late as the end of the 1990s; I had been under the impression that IDE drives eliminated the low-level formatting option (though the entry point might have persisted in BIOS for a while). I do recall that there were warnings saying not to attempt to low-level format the "new" IDE drives, because the integrated controller might respond differently to the commands from BIOS -- for instance, potentially wiping out the servo tracks that some early voice-coil-seek drives used to ensure consistent head positioning (servo track drives usually had an odd number of heads in their BIOS settings, as I recall, because one side of one platter carried the servo tracks).

Companies sinking because of bad drives wasn't new when Quantum and Maxtor did it. The first hard disk I owned was only $20 for 20 MB in 1987 because it came from Computer Memory Inc., a company which had sunk because they shipped a bunch of drives to IBM that would sometimes, without warning, consign your data to the bit bucket. That tendency was why I religiously backed up, low level formatted, and restored that drive every three months for the four years I had it. Good thing I got a deal on a bunch of slightly used 5 1/4" floppy disks formerly bulk duplicated with the previous year's tax software (and later picked up a used 720k drive in that size from a scrapped DEC Rainbow).