How many years to keep HDDs inside NAS before preventive replacement?

[emece67]

.

[james_s]

The proper approach is to keep backups of your data. Replace drives when they fail, begin to malfunction or become technologically obsolete. Don't swap out working drives just because they are old, drive failure can occur at any age though failure rates are highest in very new and very old drives.

[rdl]

Yeah, just make sure you have a good back up plan and run the drives till they fail.

I used to keep everything in triplicate (three disks) and rotated one drive out every couple of years. I now have a pile of no longer in use drives.

[barnacle2k]

My experience of the past 15 Years of operating a NAS the failure rate goes through the roof after 5 Years.

[Halcyon]

There is no hard and fast rule. It depends on the drive, the environment, how much the drives get used and sometimes a little bit of luck.

I've had 8x Hitachi Ultrastar drives running in my NAS 24/7 for the past 9 or 10 years and they are still working perfectly fine. One of them developed a bad sector a few years ago but the problem hasn't impacted data or become worse so I haven't replaced it yet.

I use ZFS which protects against data corruption and do regular scrubs of the data.

Some might say I'm living on the edge a bit, but I'm basically waiting for the cost of higher capacity SSDs to come down before I replace my array.

[nctnico]

There is no hard and fast rule. It depends on the drive, the environment, how much the drives get used and sometimes a little bit of luck.

I agree. My experience over several decades is that the durability of a hard drive depends on the operating temperature. Keep it hot and a hard drive can fail within a year. Cool it properly and it will last for over a decade easely.

I have a NAS as well and I let it report the SMART status to me daily. If the drive says it's health starts to deteriorate it is time to replace it. Like the others I strongly recommend to have a backup of all data either way.

[0culus]

There is no hard and fast rule. It depends on the drive, the environment, how much the drives get used and sometimes a little bit of luck.

I agree. My experience over several decades is that the durability of a hard drive depends on the operating temperature. Keep it hot and a hard drive can fail within a year. Cool it properly and it will last for over a decade easely.

I have a NAS as well and I let it report the SMART status to me daily. If the drive says it's health starts to deteriorate it is time to replace it. Like the others I strongly recommend to have a backup of all data either way.

Can't find the citation, but Intel did some experiments with full on hot server rooms (NOT hot aisle...whole room is kept hot) some years ago. They quickly found that the storage arrays had to be kept in a cooled room.

[Mr. Scram]

There is no hard and fast rule. It depends on the drive, the environment, how much the drives get used and sometimes a little bit of luck.

I agree. In my experience over several decades is that the durability of a hard drive depends on the operating temperature. Keep it hot and a hard drive can fail within a year. Cool it properly and it will last for over a decade easely.

Numbers from Google apparently show temperature isn't very important until you get into silly territory.

"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 "one of our key findings has been the lack of a consistent pattern of higher failure rates for higher temperature drives or for those drives at higher utilization levels.".[15] 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).[14] The correlation between manufacturers, models and the 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,[14] though it has been revealed that Google uses Hitachi Deskstar drives in some of its servers.[16]"

https://en.m.wikipedia.org/wiki/Hard_disk_drive_failure#Metrics_of_failures

[edavid]

There's almost no publicly available data to provide guidance on this.  The 2007 Google study indicated that you can't depend on SMART data to provide a pre-failure warning.  Beyond that, all we have are the Backblaze statistics.

https://www.backblaze.com/blog/backblaze-hard-drive-stats-q3-2019/

[Halcyon]

I would also take Google's figures with a grain of salt. Electronics (generally) don't like heat. We're now in 2020 and hard drive tolerances keep getting tighter and tighter.

[james_s]

I remember reading a study that suggested excessively high or excessively low temperatures increase failure rates. It makes sense to me, a hard drive is a mechanical device and mechanical parts expand and contract with temperature. They will be designed to operate optimally over a range typical of their operating environment which in most cases is going to be normal room temperature. It's similar to a car engine in that regard, you don't want want it to overheat but you don't want it to run too cold either, that's why there is a thermostat in the cooling system to allow it to warm up to normal operating temperature and regulate it there. If cool is good that doesn't mean colder is better.

I've often heard it said that electronics don't like heat, but I don't see a lot of data to back this up. Yes excessive heat will accelerate the failure of electrolytic capacitors and some other parts but that has to be *really* hot. Plenty of electronic components have no trouble operating for decades at a temperature too hot to put your hand on. I've worked in labs that were absolutely frigid with everyone wearing winter coats in the middle of summer and frankly that's stupid. Having a large number of computers in a room will obviously require significant cooling, but the temperature of the room doesn't have to be uncomfortably cold to humans, normal room temperature is just fine.

[emece67]

.

[nctnico]

"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 "one of our key findings has been the lack of a consistent pattern of higher failure rates for higher temperature drives or for those drives at higher utilization levels.".[15] 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),

Did it occur to you that a report with results going against everything that is standard in lifetime expectancy calculations (MTBF) might be wrong?

[rdl]

I doubt any of those reports or studies matter very much to the average person. The number of drives you own or ever will own is likely too small for any correlation.

[Mr. Scram]

Did it occur to you that a report with results going against everything that is standard in lifetime expectancy calculations (MTBF) might be wrong?

That's discussed a few paragraphs later. It turns out MTBF isn't all that either. That's not very surprising considering the many assumptions and variables involved. If anyone has any other numbers on how and whether MTBF and actual reliability correlate those would be most welcome.

"The mean time between failures (MTBF) of SATA drives is usually specified to be about 1.2 million hours (some drives such as Western Digital Raptor have rated 1.4 million hours MTBF),[20] while SAS/FC drives are rated for upwards of 1.6 million hours.[21] However, independent research indicates that MTBF is not a reliable estimate of a drive's longevity (service life).[22] MTBF is conducted in laboratory environments in test chambers and is an important metric to determine the quality of a disk drive, but is designed to only measure the relatively constant failure rate over the service life of the drive (the middle of the "bathtub curve") before final wear-out phase.[18][23][24] A more interpretable, but equivalent, metric to MTBF is annualized failure rate (AFR). AFR is the percentage of drive failures expected per year. Both AFR and MTBF tend to measure reliability only in the initial part of the life of a hard disk drive thereby understating the real probability of failure of a used drive.[25]"

[nctnico]

The biggest hint towards the research not being complete in the text you quoted earlier is the limited temperature range. In a well cooled server rack drives never reach very high temperatures >60 degrees C. Put a hard drive in a case / enclosure with little or no airflow and the temperatures will get into the range where the temperature becomes the driving factor of the life expectancy of a hard drive.

[Mr. Scram]

The biggest hint towards the research not being complete in the text you quoted earlier is the limited temperature range. In a well cooled server rack drives never reach very high temperatures >60 degrees C. Put a hard drive in a case / enclosure with little or no airflow and the temperatures will get into the range where the temperature becomes the driving factor of the life expectancy of a hard drive.

It's hard to get a hard drive up to those temperatures even in ill-advised scenarios due to convection and heat transfer through the supporting brackets. It wouldn't surprise me if you'd need insulation or have some hot part in close proximity. It should be noted that Google also found that temperatures below 25 degrees Celsius are detrimental to drive longevity.

[edavid]

The biggest hint towards the research not being complete in the text you quoted earlier is the limited temperature range. In a well cooled server rack drives never reach very high temperatures >60 degrees C. Put a hard drive in a case / enclosure with little or no airflow and the temperatures will get into the range where the temperature becomes the driving factor of the life expectancy of a hard drive.

I checked a couple of consumer external drives I use, and the maximum temperatures were under 50C.  OP said he was using NAS enclosures, so his drives are probably closer to 45C.

[Haenk]

Having sold quite a number of HDDs to customers over more than 2 decades and from my own experience, there are quite a few factors :
- bad batches of otherwise reliable series
- bad series/construction (late Micropolis, some "Deathstar"-IBMs or certain Seagate models)
- bad firmware (again Seagate, had that issue with early SSDs as well)
 
Since you won't know, when they fail, it's best to have that solved otherwise, like ZFS or RAID5 (or 6 or like I did with our main server as RAID50)

[nctnico]

The biggest hint towards the research not being complete in the text you quoted earlier is the limited temperature range. In a well cooled server rack drives never reach very high temperatures >60 degrees C. Put a hard drive in a case / enclosure with little or no airflow and the temperatures will get into the range where the temperature becomes the driving factor of the life expectancy of a hard drive.

It's hard to get a hard drive up to those temperatures even in ill-advised scenarios due to convection and heat transfer through the supporting brackets. It wouldn't surprise me if you'd need insulation or have some hot part in close proximity.

Well, you should see at what kind of less optimal places (to put it mildly) people place computers & associated hardware.

It should be noted that Google also found that temperatures below 25 degrees Celsius are detrimental to drive longevity.

That is an interesting find indeed.

[Rerouter]

If your device supports it, setup an email or other critical notification if anything changes about any of the critical SMART data, I use hard disk sentinel for this,

A hard drive that is happily spinning away and not showing any issues via SMART is technically more trusted than a brand new untested drive, as a drive that has been running for years is less likely to face a critical fault if disturbed, vs a new drive that could have manufacturing faults

To this end, any new drive I will tend to run 3-5 verified erase passes with random data, So far it has caught all early faults,

If you wish to verify that your drives are fully healthy, make the device perform a full volume read test, while most enterpise like weekly, I prefer monthly at a low transfer rate. (Raid card handles this for me)

Other small thing is a lot of SSD's with many reallocated sectors, but low wear leveling can be fine to keep using, just need to force it to read over the drive a few times and prompt the controller to reallocate everything, you will usually loose some space on the drive, but after that bad area is dealt with, the rest is usually ok.

[Haenk]

It's hard to get a hard drive up to those temperatures even in ill-advised scenarios due to convection and heat transfer through the supporting brackets. It wouldn't surprise me if you'd need insulation or have some hot part in close proximity.

A certain enclosure we sold (I'd guess about 10k of them) was essentially a closed metal frame inside a closed acrylic frame. 7200rpm drives ran really hot in there, easily reaching 60oC or more. However this hasn't led to a higher failure rate (compared to drives without enclosure), surprisingly.

[Mr. Scram]

A certain enclosure we sold (I'd guess about 10k of them) was essentially a closed metal frame inside a closed acrylic frame. 7200rpm drives ran really hot in there, easily reaching 60oC or more. However this hasn't led to a higher failure rate (compared to drives without enclosure), surprisingly.

A closed acrylic frame with zero convection is nigh on intentionally insulating the drive. That's probably a worst case scenario. Your experience seems to mirror what the big hard drive consumers report though.