Storage

10 ways to keep hard drives from failing

With a little proactive care, you can extend hard drive life, squeezing more value from your company's equipment investments, reducing user downtime, and preventing catastrophic loss of data.

Hardware prices have dropped considerably over the last decade, but it's irresponsible not to care for the hardware installed on machines. This is especially true for hard drives. Hard drives are precious commodities that hold the data employees use to do their jobs, so they should be given the best of care. Inevitably, those drives will die. But you can take steps to prevent a premature hard disk death. Let's examine 10 such steps to care for the health of your drives.

Note: This article is also available as a PDF download.

1: Run chkdsk

Hard disks are eventually going to contain errors. These errors can come in the shape of physical problems, software issues, partition table issues, and more. The Windows chkdsk program will attempt to handle any problems, such as bad sectors, lost clusters, cross-linked files, and/or directory errors. These errors can quickly lead to an unbootable drive, which will lead to downtime for the end user. The best way I have found to take advantage of chkdsk is to have it run at next boot with the command chkdsk X: /f where X is the drive you want to check. This command will inform you the disk is locked and will ask you if you want to run chkdsk the next time the system restarts. Select Y to allow this action.

2: Add a monitor

Plenty of applications out there will monitor the health of your drives. These monitors offer a host of features that run the gamut. In my opinion, one of the best choices is the Acronis Drive Monitor, a free tool that will monitor everything from hard drive temperature to percentage of free space (and everything in between). ADM can be set up to send out email alerts if something is amiss on the drive being monitored. Getting these alerts is a simple way to remain proactive in the fight against drive failure.

3: Separate OS install from user data

With the Linux operating system, I almost always separate the user's home directories (~/) from the OS installation onto different drives. Doing this ensures the drive the OS is installed upon will enjoy less reading/writing because so much of the I/O will happen on the user's home drive. Doing this will easily extend the life of the drive the OS is installed on, as well as allow you to transfer the user data easily should an OS drive fail.

4: Be careful about the surrounding environment

Although this seems like it should go without saying, it often doesn't. On a daily basis, I see PCs stuck in tiny cabinets with zero circulation. Obviously, those machines always run hot, thus shortening the lifespan of the internal components. Instead of shoving those machines into tight, unventilated spaces, give them plenty of breathing room. If you must cram a machine into a tight space, at least give it ventilation and even add a fan to pull out that stale, warm air generated by the PC. There's a reason why so much time and money have gone into PC cooling and why we have things like liquid cooling and powerful cooling systems for data centers.

5: Watch out for static

Here's another issue that should go without saying. Static electricity is the enemy of computer components. When you handle them, make sure you ground yourself first. This is especially true in the winter months or in areas of drier air. If you seem to get shocked every time you touch something, that's a good sign that you must use extra caution when handling those drives. This also goes for where you set those drives down. I have actually witnessed users placing drives on stereo speakers, TVs, and other appliances/devices that can give off an electromagnetic wave. Granted, most of these appliances have magnets that are not strong enough to erase a drive. But it's a chance no one should take.

6: Defragment that drive

A fragmented drive is a drive being pushed to work harder than it should. All hard drives should be used in their most efficient states to avoid excess wear and tear. This includes defragmenting. To be on the safe side, set your PC(s) to automatically defrag on a weekly basis. This works to extend the life of your drive by keeping the file structure more compact, so the read heads are not moving as much or as often.

7: Go with a solid state drive

Solid state drives are, for all intents and purposes, just large flash drives, so they have no moving parts. Without moving parts, the life of the drive (as a whole) is naturally going to be longer than it would if the drive included read heads, platters, and bearings. Although these drives will cost more up front, they will save you money in the long run by offering a longer lifespan. That means less likelihood of drive failure, which will cause downtime as data is recovered and transferred.

8: Take advantage of power save

On nearly every OS, you can configure your hard drive to spin down after a given time. In some older iterations of operating systems, drives would spin 24/7 -- which would drastically reduce the lifespan of a drive. By default, Windows 7 uses the Balanced Power Savings plan, which will turn off the hard drive after 20 minutes of inactivity. Even if you change that by a few minutes, you are adding life to your hard drive. Just make sure you don't shrink that number to the point where your drive is going to sleep frequently throughout the day. If you are prone to take five- to 10-minute breaks often, consider lowering that time to no less than 15 minutes. When the drive goes to sleep, the drive is not spinning. When the drive is not spinning, entropy is not working on that drive as quickly.

9: Tighten those screws

Loose mounting screws (which secure the hard drive to the PC chassis) can cause excessive vibrations. Those vibrations can damage to the platters of a standard hard disk. If you hear vibrations coming from within your PC, open it and make sure the screws securing the drive to the mounting platform are tight. If they aren't, tighten them. Keeping your hardware nice and tight will help extend the life of that hardware.

10: Back up

Eventually, that drive will fail. No matter how careful you are, no matter how many steps you take to prevent failure, the drive will, in the end, die a painful death. If you have solid backups, at least the transition from one drive to another will be painless. And by using a backup solution such as Acronis Universal Restore, you can transfer a machine image from one piece of hardware to another piece of hardware with very little issue.

Other suggestions?

Have you found any of the above steps to be especially helpful in extending the life of a hard drive? What other form of wisdom or magic keeps those disks from failing prematurely? Is there a brand you have found that lasts longer than others? Share your experiences with your fellow TechRepublic readers.

About

Jack Wallen is an award-winning writer for TechRepublic and Linux.com. He’s an avid promoter of open source and the voice of The Android Expert. For more news about Jack Wallen, visit his website getjackd.net.

36 comments
msawyer91
msawyer91

There are a lot of good hard disk health tools out there, and some monitor SSD health too. I'm partial to WindowSMART 2013 myself (http://www.dojonorthsoftware.net/WindowSMART.html) which monitors the health of HDDs and SSDs, and shoots you an email as well as an alert on your iPhone, Android or Windows Phone mobile device. It'll even back up the data--files and folders you select ahead of time--from the disk. The backup is nice, because it backs up only what you consider to be important. Why back up the Program Files folder? Chances are your important stuff are your pictures, videos and documents, so those should be your priority. WindowSMART 2013 works with all versions of Windows, from XP SP-3 right through Windows 8, including the Server editions. If you have hardware RAID, then you should look at HD Sentinel - http://www.hdsentinel.com The good thing about both WindowSMART and HD Sentinel is that they are PROACTIVE tools, so they always keep you in the know about your disks. Most HDDs and SSDs give advance warning that trouble is afoot, so why not know about the trouble when there's time to do something about it? That makes WindowSMART and HD Sentinel not just hardware tools, but cost cutting tools.

satance
satance

What do you think, discs have to use power-saving facility or it's better they are running allways? As my knowledge when they run surfaces are free of powder but in other hand motor is getting older, what to consider as fundamental causes of disc death?

fletchoid
fletchoid

I like to clone my system drive every few months using Acronis True Image. I actually have a "Master Clone" which is just the OS, drivers, and some core programs like antivirus, office programs, antimalware programs and some utilities. I also have a "working" clone, which contains everything. Now, when a hard drive goes stupid on me, I slap in a clone, and am ready to go. Sure, I may have to reinstall a few programs, but that is better than reinstalling the OS, drivers, antivirus, etc. etc. I also keep all data (photos, music, documents, etc) on a separate drive and back them up frequently. When you can buy a 500Gb drive for 40 bucks, why not have some clones to back you up?

Kent Lion
Kent Lion

Something related to but possibly more important than "4: Be careful about the surrounding environment": When a fan moves air in or out of the box, that air must come in or go out somewhere else. I have opened up PCs to find huge chunks of dust effectively plugging those inlets/outlets (and sometimes the fan's screen, too). People need to open that box and carefully use a vacuum cleaner on it (but ground the box and be careful not to get the nozzle too close to any electronics, moving air generates static electricity).

rpnadal08
rpnadal08

Thou good info, but, there is no way to keep a hard drive from failing. It's a mechanical device with a circuit board that's very sensitive to voltage spikes. Everything mention in your article is great for the smooth operation of Windows. There are steps one can take to reduce failure and the main one is keeping the inside of the computer clean. I have seen PC's come to me that when opened they have a foot of dust on the board, animal fur, food crumbs, mice droppings and things that have no name. I have taught all my clients on how to clean their PC's and how extend the life of their equipment. After all this said, it's still a mechanical device! I saw them last 30 seconds out of the box to 20 yrs and still working. Static electricity is another hard drive killer, rugs and computers do not mix, a piece of wood under the unit will do wonders for the PC.

NickNielsen
NickNielsen

Reducing the chance of static discharge is about reducing the difference in potential between you and your work. The easiest way to do this when working on a computer is to touch the chassis and remain in contact as much as you can when installing/replacing components. The new power supplies that remain hot when the computer is off make it very dangerous to actually connect yourself to ground when working on the PC.

jmbrasfield
jmbrasfield

7: Go with a solid state drive Though I admit solid state drives have improved, I have not had a hard drive fail in the last 10 years. My present system has two Hitachi 250G EIDE DeskStar drives (ten years old +/-) and two Hitachi 250G SATA DeskStar drives (five years old +/-) and never an error. I can not same the same for any solid state drive I have installed for customers in the last year or so, now that they have become so popular for laptops and netbooks. For now I will stay with old school, or at least I should say Hitachi DeskStar old school.

vindasel
vindasel

Heat and vibration are the worst enemies of a mechanical HDD. Avoid these two problems, and half the battle is won already. Regarding putting the drive into power save: I was under the impression that the drive experiences maximum mechanical stress when it starts to spin up and spin down (acceleration and deceleration respectively). So it might actually be a good idea to let it run continuously rather than start up/shut down frequently. Can someone confirm this? I agree with the rest, especially chkdsk and defragmentation. Chkdsk is an underrated but often useful tool to check for filesystem errors. Ofcourse, a defragmented drive has to do less work in the long term to read/write contiguous files compared to fragmented files, so that's always a good thing. Just stick a reliable automatic defragmenter on Windows rather than relying on the native utility, and you're good to go. As for SSDs, sure, they're more reliable than mechanical HDDs in some ways, but if they experience material breakdown (crystal structure defects etc at the cell level) you can still lose the drive if I am not wrong.

styman60
styman60

i think these are all good ideas

dennis
dennis

I'm surprised no one has mentioned keeping drives cool. Especially in a server, NAS, SAN device, use a case or dirve bay that is designed to direct air flow over the drives.

fbenitez
fbenitez

Solid State Drives are super fast.

pbkeith
pbkeith

Good article and an important topic to discuss. For #4, I might add that mobile users should be trained to either shutdown or hibernate laptops before putting them in their bags. I frequently see users with failed hard drives, who have literally cooked them by putting their laptops in standby when going to and from work or when traveling. Standby, though it is a low-power mode, is still powered on, and still generates heat.

Slayer_
Slayer_

The powering up and down of the drive causes more ware and tear than just leaving it spinning all the time. On a New drive, let it power down, but as the drive is getting very old, leave it on. I have some 1gig drives that are still working perfectly because we leave them spinning all the time. On my home PC, my primary drive is now 8 years old, when you fire it up and its cold, it whines like crazy but after it warms up, its good. I don't let that drive turn off very often, it has like 2 hours before it turns off. My Win98's 10gig Fuji drive (Those that all failed) is still going strong after all these years because its never stops moving. When the computer does get turned off, the spin up time on that drive is almost 30 seconds. (no critical data on that drive obviously)

sura.jan
sura.jan

Chkdsk can be very dangerous! Before running chkdsk run Spinrite to renew all data on the disk and to be sure that problems are really on the disk! Before running chkdsk with corrections be sure that problems with reading are on the disk and not in power or signal cables! With bad power cable connector chkdsk tries to repair data and so it destroys them (you receive useless file "fragments" instead of data! The disk itself can be OK and you loose data.

danwdoo
danwdoo

Most of the steps aren't about preventing a drive from failing, almost nothing can do that, but about how to maintain your drive and how to prepare for when it does fail. #7 - hugh? Solid state memory has a limited number of writes before failure. While they are getting better, they still have fairly high failure rates. See: http://www.computerworld.com.au/article/371175/solid_state_drives_no_better_than_others_survey_says/?fpid=1 plus any number of other articles out there discuss this issue. When you add to this the cost premium plus the reduced capacities, SSD's aren't looking so good unless you just want battery savings or speed.

NickJames999
NickJames999

One of the biggest and most consistent causes of failure for disks and electronics is the transport of equipment on hard trolleys with hard tyres over rough surfaces with no foam or other protection. Usually this mistake is committed by the junior staff that get the 'ship it around' job most. Compare this with the care manufacturers take when they ship it to you. All that packaging has a reason as well as a cost.

alexisgarcia72
alexisgarcia72

In regards to SSD reliability, there is several different quality builds. I have a couple of SSD disks for a server with 7 years of warranty!

Netweezurd
Netweezurd

Point 9 says to tighten up the screws to prevent vibrations... ya... ok... The real reason for the 4 screws is that the metal case is the actual heat sink of the drive. A lot more heat gets suck off the drive casing by the metal frame that holds it in place. Personally, I stay away from plastic sliders. What better way to sell drives, make the industry turn, make a case that the drives aren't cooled properly. Metal sliders... i dunno yet. MotherDawg I do RPMs I own crappy boxes GNU/Linux 2.6.34.7-63.fc13.i686

Oz_Media
Oz_Media

SS drives are fast but prone to failure and are still very small in comparison. I'm sure in 10 years that'll change but they still have a long way to go before they offer a great benefit for anything other than portable devices.

travis.duffy
travis.duffy

If that were true, drives in SAN/NAS devices would have more failures as they spin constantly. But they do not.

TelcoChuck
TelcoChuck

I especially agree with running Spinrite perodically to find surface errors. There might be another program that does the same thing, but I've used Spinrite for a while and it does seem to clean / refresh the drive surfaces. No, I don't work for them, I just like the results.

Excelmann
Excelmann

Good, bad, pros, cons, adequate? Does any empirical evidence exist? Are there other 32-bit and/or 64-bit products which are clearly superior?

twaynesdomain-22354355019875063839220739305988
twaynesdomain-22354355019875063839220739305988

+1! I agree with you. This article seems to have been written with almost no forethought or consideration for reality. This kind of misinformation does no good, in fact can do harm to, newbies and those just reaching the lower experienced levels. If bias was your intention, you have caused me to be biased against you; a partial win for you.

Tony_Scarpelli
Tony_Scarpelli

I tell all my clients they need their heads examined if they allow there important data and processos to be on hard drives older than 3 years. When I got into IT a PC was $2700. Today you can buy a whole small network for $2700. If your business isn't worth a new hd now and then or a new pc/workstation/server then you pretty much just waiting for some event to kill your info and business. SBA did a study 60% of small (under 100 employees) companies who had a catastrophic data failure went out of business within 3 years.

Realvdude
Realvdude

So, without any evidence to support my statement, I'd say stripe two drives so that they only have to handle half the reads and writes of a single drive. Seriously though, when I recently looked at drives specs again, I noticed a rate for spin-ups or another term for the drive coming out of a sleep state. That made me wonder if power saving mode is even good for drive life. I think that manufacturing quality has a lot to do with drive failure rates. I've got a server that has been running 247 for over 4 years with two 80Gb drives mirrored. The ECC memory has actually failed, though it is a little older than the drives. Neither drive is rated for server use. The server is running 2000SBS w/Exchange and SQL Server for a company of around 10 employees. FYI - Another interesting bit I read about drives recently, had to to do with RAID. It warned about the possibility of unrecoverable data loss, due to the use of the same drive for all the drives in a RAID. This makes sense due to the fact that drives of the same model and date of manufacture, running in the same environment, will likely fail about the same time. Kudos to Jack for getting us all thinking about the topic.

alexisgarcia72
alexisgarcia72

I have a a couple of buffalo NAS with 4 drives, spinning all the time. Zero failures. I have a Pillar SAN with 20 hard drives, 5 years old, SATA drives, Zero failures. The Small NAS WD with power saving features (in idle, you hear how the HD stops spinning) is more propense to failure.

Oz_Media
Oz_Media

I have used Win7 for JUST under a year now, if not a year right about now, and I have never NEEDED to defrag it. I went through the system the other day, cleaning registry, orphaned files, bad uninstallers etc. and went to defrag, after the HD scan finished it was at less than 1% fragmented, a YEAR into it. I add and remove hoards of programs ,music, temp files for work in progress, videos, games etc. Several gigs each day added to, copied, moved, deleted etc. less than 1% in a year! DEFRAG? I just haven't found a need or use in Win7 yet. When I firt got this notebook, I used teh auto defrag and never noticed a single blip or intertuption if it ran while I was working, I didn't even notice it running most of the time. When I shut off autodefrag, there was no speed increase and now many months later still no need to defrag. Win7 handles memory, HD and processes SO much better than anything they've done before that popular releases like XP don't even come close to standing up to it. I've never been a Windows (MS) fan, in fact I purposely chose to cert in Novell as I hated MS all around. I simply cannot deny that Win7 is a good platform though, it just keeps going like the Energizer bunny, it seems thay finally got it right this time.

james
james

I've been happy with JkDefrag for awhile. Has a number of handy features...

rwtodd2007
rwtodd2007

Works best for our users and my home computers, and its free.

Slayer_
Slayer_

Customers wanted everything cheap. So I built a backup system that ran every night and made an exact copy of their data, one in the cloud, and one onto a USB drive. Then I just checked the SMART data on the drive occasionally and if it looked like it was going to fail, then replaced it. only 1 drive started failing after a year, the others last (far as I know) 6 years or more. Not my customers anymore after 6 years, got myself a proper job programming.

fletchoid
fletchoid

One lousy bit fails on either drive and your whole system is pooched. Better to "mirror" your drives, so if one fails, slap in a new one and on you go.

john3347
john3347

"FYI - Another interesting bit I read about drives recently, had to to do with RAID. It warned about the possibility of unrecoverable data loss, due to the use of the same drive for all the drives in a RAID. This makes sense due to the fact that drives of the same model and date of manufacture, running in the same environment, will likely fail about the same time." I would like to see the results of some long term testing of this phenomenon. I believe the failure of the drives discussed in this statement would be as random as soldered-in transistors of twenty or thirty years ago where either of two transistors of the same manufacture, same manufacture date, same operating conditions, etc. may have an actual service life of many times (perhaps 20 or 50 times) that of another. I don't think that because one "conventional" harddrive fails, that it is a prediction that a twin to it is on the verge of immediate failure. It is true, however, that a mechanical harddrive will eventually fail......it may fail in 4 days, or it may fail only after 38 years. The life of a solid state drive may be predicted to a closer estimate because it is known to fail after some finite number of write-rewrite cycles. In certain raid configurations, the failure of one solid state drive MAY be an accurate predictor of the imminent failure of its twin.

twaynesdomain-22354355019875063839220739305988
twaynesdomain-22354355019875063839220739305988

I disagree that two drives of the same model and date of manufacture will fail simultaneously, or nearly so. You won't get anything of the kind because the components IN the drive are not of the same age or date of manufacture and only a very statistically unrealistic chance of them being consecutively manufactured. The DRIVE info has nothing to do with when the internal components were manufactured or even are the same "model". They would even be very unlikely to have the same failure mode! In addition, striped drives aren't much different than any other: They do NOT each get the SAME data or amount of data when they're written to. My 2 cents anyway. Twayne` Remember your curves from statistics?

dennis
dennis

I disagree with the comment about using the same make and model drives in a RAID, that they would fail at nearly the same time. The likelihood is extremely low, and I'd rather match drive performace in a RAID. I have never seen it happen among the 100 or so RAIDs I've admin'ed over the years. The part I agree on is, if it is reasonably possible to use same model drives of different dates or batches, that would reduce the odds more.

alexisgarcia72
alexisgarcia72

There is no warranty on earth to cover data. This is why we perform full backups every monday and differential backups the remaining days of the week. And BTW, add to this, we have fresh images with Acronis so we can restore a damaged HD in minutes (and we have Vmware vmotion ready in place).... thanks!!