We have a lot of PCs in the company and nobody wants to wipe a multitude of hard drives. We also have many apprentice toolmakers who really want to destroy things. Thus, every couple of months, our apprentices receive two heavy baskets of hard drives to drill through.

Some of my coworkers believe that this is absolutely overkill. I, however, believe that not wiping the drives before drilling through them might make some data recoverable.

According to this question, wiping with DBAN will make data completely unrecoverable.

DBAN is just fine. Here's the dirty little secret--any program that overwrites every byte of the drive will have wiped everything permanently. You don't need to do multiple passes with different write patterns, etc.

How about drilling a hole?

  • 26
    Always encrypt everything on your hard drives, even though it is an hassle during reboots. This way you are sure no-one can read the hard drive, even in the case of leaking a hard drive by accident – Ferrybig Aug 16 '17 at 7:31
  • 12
  • 26
    If they're enthusiastic, have them use a belt sander rather than a drill. – Mark Aug 16 '17 at 18:43
  • 22
    Why not just get a metal forge. Put all the metal in heat to whatever thousand degree, and let melt. Your apprentice toolmakers can make something new with the metal. Now the data is destroyed. – cybernard Aug 16 '17 at 19:00
  • 13
    Nuke them from orbit. It's the only way to be sure. – Zenadix Aug 20 '17 at 1:39

17 Answers 17

up vote 161 down vote accepted

Drilling a hole in the drive enclosure which passes through all the platters will make it impossible to run the drive. Most modern HDDs don't have air inside the enclosure, and you've let what was in there escape. You've filled the cavity with tiny pieces of drill swarf, which will be on everything including the platters, and will crash the heads if someone tries to lower them onto the rotating platters. You've also unbalanced the platters, though I don't have an estimate for whether this will be fatal. The drill bit will likely pass through the controller board on the way, which though not fatal will certainly not help anyone trying to hook the drive up.

You have not prevented someone from putting the platter under a magnetic force microscope and reading most of the data off that way. We can be fairly sure this is possible, because the SANS paper linked from the linked SF article demonstrates that you can't recover data from a platter with an MFM after a single overwriting pass, and such a test would be completely meaningless if you couldn't recover non-overwritten data using the same procedure.

So drilling through the platters will very likely prevent data from being read off the HDD by normal means. It won't prevent much of the data being recoverable by a determined, well-funded opponent.

All security is meaningless without a threat model. So decide what you're securing against. If you're worried about someone hooking up your old company HDDs and reading them, after they found them on ebay / the local rubbish dump / the WEEE recycling bin, then drilling is good. Against state-level actors, drilling is probably insufficient. If it helps, I drill most of my old drives, too, because I am worried about casual data leakage, but I doubt the security services are interested in most of my data. For the few drives I have which hold data that Simply Must Not Leak, I encrypt them using passphrases of known strength, and drill them at the end of their lives.

  • 13
    Most HDDs have their board (and arm "engine") off the disks. That's why they are rectangular and not square. You won't hurt the board by drilling the disk (you may destroy the head's arm though). Of course HDDs have air inside (otherwise you couldn't not open them without excessive force) they even have breathing hole with filter. Open one and you will see. A determined attacker could be able to open the disk clear, replace damaged heads if needed, rebuild and read (most of) the data. – goteguru Aug 16 '17 at 9:30
  • 16
    @goteguru "Of course HDDs have air inside" Except those that don't, of course; such as for example these new-fangled helium-filled drives. – a CVn Aug 16 '17 at 11:36
  • 13
    @MichaelKjörling Yes, yes of course. All has some gas in them. That's the correct wording. Nevertheless the point is the same. Helium filled drives will go happily with simple air. No problem. Dust, maybe humidity can kill the drive fast, special gas is just an extra. – goteguru Aug 16 '17 at 11:52
  • 11
    A while ago, I visited the Information Security Research Group's Computer Forensics Research Lab and the Senior Lecturer / Advanced Data Recovery Consultant demonstrated the many ways that they can recover data off of HDDs and said that, pretty much, the best way to destroy HDDs and their data is to leave the platters in water then air to rust them. – mythofechelon Aug 16 '17 at 15:34
  • 34
    Upvoted specifically for this... All security is meaningless without a threat model – barbecue Aug 18 '17 at 18:23

The security policy for many companies is to universally physically destroy all data carriers, so plain old paper documents and prints, spinning hard disks, SSD's etc. all get shredded before they get recycled.

In that regard your question might be irrelevant and you may simply need to comply to that policy.

With SSD's becoming more prevalent it is also good to realise that software wipes are not reliable for SSD's.

With regards to physically destroying drives by drilling a hole: That will prevent normal usage, resale and refurbishing.

In many cases that may be sufficient, but while drilling a hole makes the disk inoperable that still only destroys a fraction of the data. With sufficient money to spend a determined attacker can still recover the remaining data. If that is a risk is something you need to determine for yourselves.

  • 9
    +1 from me for the SSD case, plus the pointer about policy. +another, if I could, for drilling a hole ... will prevent ... resale: I'm now imagining the worst eBay feedback evar. – MadHatter Aug 16 '17 at 7:05
  • 4
    @MadHatter it happens frequently enough that security researchers buy batches of old drives, laptops and workstations and recover sensitive data. I can also quite easily imagine an engineer otherwise skipping the intermediate steps of binning gear and dumpster diving after following the companies policy to "dispose of old drives" and simply taking them home and putting them up on eBay for a quick buck. – HBruijn Aug 16 '17 at 7:22
  • 2
    Yes, indeed, see "found them on eBay" part of my answer above. – MadHatter Aug 16 '17 at 7:35
  • 4
    +1. for spinning drives, overwriting all sectors renders the data impossible to recover even for determined attackers, AND avoids wasting perfectly fine hard drives. – JanErikGunnar Aug 16 '17 at 10:43
  • 4
    Note that an SSD is comprised of multiple chips - several storage chips running in parallel connected to some controller chips. Depending on where you drill the hole, and how many holes you drill, the data may be more or less recoverable. I think all in all, running the TRIM command on the entire SSD is more secure than drilling it as well. – JanErikGunnar Aug 16 '17 at 10:48

Don't drill all the way through, just through the top of the housing. Pour in thermite and ignite!*

  • Definitely safer than drilling one hole all the way through.
  • Probably a lot safer than overwriting every bit too.
  • This will even take care of SSDs, though they may not have a hollow for powder to fill.
  • Your apprentice toolmakers will think this is a lot more fun even than drilling!

*do this outside.

  • 29
    Depending on what the apprentices are apprenticed as, "safer" is a relative concept when igniting thermite ;) – Toby Aug 16 '17 at 11:40
  • 8
    Related youtube.com/watch?v=-bpX8YvNg6Y :) ... Trying to "wipe" it with thermite, while it probably looks pretty cool, might not be the most feasable way to do it. – rinukkusu Aug 16 '17 at 12:25
  • 4
    This technique completely defeats the purpose of drilling. If you have time for this, you have time to simply wipe the drives. – Dmitry Grigoryev Aug 16 '17 at 14:42
  • 4
    "safer" is definitely not the correct word to use here. do you mean more secure? in both your first and second points. how is overwriting bits not a safe thing to do? how is drilling a hole not safe? – Octopus Aug 16 '17 at 18:45
  • 2
    Or in my case, I decommissioned a cell-phone battery at the same time — two items killed with one blow (or nail, as the case may be). – JDługosz Aug 17 '17 at 8:52

While drilling a hole is sufficient against most real-life attackers, why not buy an HDD shredder? It's only $3000 to $5000 for smaller models, and it works pretty well with SSDs too. Also, having your drives shredded will sound much more convincing in case of an audit than "we have drilled holes in them".

  • 2
    just an industrial shredder with same capability may cost less.. specialized shredders are overpriced. E.g. here, in Russia it is almost a monopoly for one company-importer to exploit. – Swift Aug 16 '17 at 14:31
  • 1
    Just wondering... If drilling a hole is sufficient against "most real-life attackers", just what class of attackers would you consider hole-drilling to not be sufficient against? – a CVn Aug 21 '17 at 15:04
  • 2
    @MichaelKjörling He did say 'most,' not 'all.' I suppose the wording also leaves open the possibility of imaginary attackers, dead attackers, or perhaps undead attackers. I'm not sure how attackers get multiplied by sqrt(-1), but the wording does seem to suggest such attackers might pose a threat. – reirab Aug 21 '17 at 15:39
  • 2
    @reirab I will readily admit that my comment was written slightly tongue in cheek, but given that the question was whether drilling a hole is sufficient to render data unrecoverable, it's not unreasonable for an answer that claims it will be for only some adversaries to discuss which class of adversaries it would not provide sufficient difficulty for. Not everyone has the luxury of not being of interest to relatively capable adversaries; if OP is considering drilling holes into HDDs to render them unusable, it stands to reason that they might at least be considering powerful adversaries. – a CVn Aug 22 '17 at 7:21
  • 1
    @MichaelKjörling Oh, yes, I agree. My comment was a bit tongue-in-cheek, too. :) If your adversary is an APT (e.g. a state-level actor,) drilling holes may very well not be sufficient. – reirab Aug 22 '17 at 14:56

Its worth remembering that drilling and other physical destruction methods are relatively fast compared to a wipe, and it is simple to verify that the disk has in fact been processed by looking at it, since unlike a wiped and unwiped disk, it is obvious that a disk with a hole in it will not work.

So, either a few hours, or a minute or (less!) with a drill press per disk.

You'd obviously want to tailor your approach for SSDs but the advantage with physical destruction for a lot of disks is speed and relative verifiability that the data on the disks is no longer recoverable.

Drilling, or disassembling the platter stack and bending/breaking the platters, will certainly make any non-laboratory, non-multi-$1000 recovery effort futile. Any HDD, even 1980s types, relies on the platter surface being perfectly level, since aerodynamic effects are used to keep head and platter very close to each other without touching. Any reading method that can deal with a bent or perforated platter no longer resembles a hard drive, and would certainly not only require expensive and/or custom made equipment but would also be orders of magnitude slower than reading from an intact drive.

Theoretically, someone could attempt to modify the drive to do a partial recovery on tracks not interrupted by the hole, so to be safe, drill several holes so most concentric tracks are effected.

Damaging or removing (and separately disposing or keeping) the circuit board is either pointless (if dealing with an even slightly determined attacker - using a replacement circuit board from the same model is a common technique in data recovery) or absolutely sufficient (to deter opportunistic, trivial attackers like someone that would resell an intact drive not meant to be sold).

BTW, any claims about software-wiped data being recoverable depends a lot on two things: a) which encoding type was used (2000 and newer will likely be PRML, which already exploits any error margins it can to store more data), b) how the wiping method deals with HBA features and fault-remapping algorithms (and the spare sectors they use) in the drive (crude wipe programs usually won't, builtin "secure erase" firmware usually will).

  • 2
    If a drive's controller encrypts data using a key that is not stored anywhere in the universe except in the controller, destroying the controller may render the drive's contents permanently unrecoverable. Depending upon circumstances, that might be viewed as a good thing or a bad thing. – supercat Aug 26 '17 at 18:39

This question reminded of something. I studied Electrical and Electronics Engineering. We had a lecturer who used to work for the army. In one lecture he said that from time to time the army destroyed some HDDs.

Asked if anyone knows the correct answer about how, many answers came. Then he said, we take a sledgehammer. Make sure that it gets beaten hard. Right after that we shred it.

My only reaction to that was "primal". Seems like it is the correct way to destroy a HDD.

Also I suggest you to read this link: https://community.spiceworks.com/topic/586771-the-leftovers-is-drilling-holes-in-an-old-hard-drive-really-enough

It will most probably answer many of your questions.

  • 6
    Shredding I like; I'm less convinced that hitting it with a sledgehammer beforehand adds much. It seems a bit like administering a punishment beating to your treacherous henchman, then shooting him in the head. The beating might make you feel better, but it seems unlikely to have any long-term effect on his behaviour. – MadHatter Aug 18 '17 at 16:16
  • 1
    When I was in the Navy, we had to destroy thousands of disk platters and magnetic tapes when a secret project was dismantled. We had degausing wands (strong magnets), and had to wave them over the reels of tape and platters in a specific manner, then remove classification markings. – Steve Aug 18 '17 at 16:34
  • 2
    The army has access to explosives. Would exploding them with handgrenades or similar be more fun than just sledgehammering? – Thorbjørn Ravn Andersen Aug 19 '17 at 11:45
  • 1
    @ThorbjørnRavnAndersen no idea, but if I was at the army I'd explode them so hard. – Tuğberk Kaan Duman Aug 19 '17 at 12:29
  • 2
    I have destroyed hard drives with a sledgehammer. Highly recommended as therapeutic. – dmourati Aug 23 '17 at 3:18

Hard drive platters are made either of aluminium or tempered glass. If you are to quickly render the data of many such drives absolutely irrecoverable, using a drill-press will utterly destroy the glass ones. (Even the smallest damage will cause them to fracture into thousands of shards.) After drilling the hole, the aluminium platters would be destroyed best / easiest by injecting a quantity of a strong lye (NaOH) solution. There are other methods f.e. throwing them into an active volcano, but this is how I would do it. 'Could do 50 over drives in an hour like that, I reckon. Do wear eye protection though -not only is lye a nasty substance, but highly tempered glass will violently shatter and project very dangerous shards.

  • 21
    Please don't put things on train tracks. – David Richerby Aug 17 '17 at 20:10
  • The fact that my "dangerous suggestion" about putting it on the tracks was in the same sentence withe the patently, even more absurd method of throwing it in an active volcano, flew clean over your head, no? – William Snoch Sep 14 '17 at 12:50
  • (Ignore my unfinished comment) The fact that my "dangerous suggestion" about putting it on the tracks was in the same sentence with the patently, even more absurd method of throwing it in an active volcano, flew clean over your head, no? I mean, folks who are so stupid not to get the jocular hyperbole of this, should rather not use dangerous power-tools, or caustic chemicals either. I suppose You would commend me for not advising to use 500 grams of Semtex as a method of destroying data on hard-drives. Not to mention cutting them up in small pieces and then eating them...really! – William Snoch Sep 14 '17 at 13:00
  • You suggested three things. One (drill press) was your recommendation. One (volcano) was so absurd that probably nobody would try it (most people don't live anywhere near an active volcano). The other one (train tracks) was of completely unclear status. It isn't obviously absurd, and the majority of the site's users probably live close enough to a railway line to try it. You say you intended it as a joke but this wasn't at all obvious, and there was a real danger that somebody would take it as a serious suggestion. Most people don't have access to semtex; eating is obviously absurd. – David Richerby Sep 14 '17 at 13:02
  • If you're using the web version of the site, you can mouse-over your unfinished comment and a grey cross icon will appear -- click that to delete the comment. If you're using the app, you can tap your comment and a bunch of options appear -- I think one of those is delete. – David Richerby Sep 14 '17 at 13:04

I take a two-pronged approach to sensitive data and old drives, that involve my children:

  • All /home/user directories are encrypted (with default Ubuntu home directory encryption).
  • All user data except for large work-in-progress media files, that is stored in custom directories, is encrypted with EnCryptFS. (Possibly doubly encrypted if under a /home/user directory - for standardization, not "extra encryption".)
  • If the drive is still operational at EOL:
    • If it contained potentially sensitive data: a one-pass randomized wipe is performed.
    • If no sensitive data: A new MBR or GPT table is written, a new partition created, and a few MB of random data is written. (Time permitting.)
  • All SSDs get hit repeatedly with a hammer and discarded.
  • All HDDs get handed to my children. They can do anything they want with them, with minimum parameters:
    • All platters must be physically, individually removed.
    • Platters must be physically scrambled/jumbled from all drives and within each drive (e.g. by scattering on the floor and shuffling them around).
    • Making duct-tape animals out of the platters and parts is totally OK.
    • Scratching the platters is totally OK.
    • I've never gone farther than that, but throwing the platters in a bonfire, or taking a torch to them, might make for a fine additional step.
  • 7
    How does this answer the question of whether drilling a hole into a hard drive makes data irrecoverable? – a CVn Aug 17 '17 at 10:56
  • 4
    Two points: 1) I meant to reply with a comment rather than an answer. [Forgive me, I'm new.] 2) Many of the other comments AND answers are lighthearted and do not directly answer the question - while still sincerely addressing the implicit larger question [that also seemed implied], with excellent ideas and points. I get that that many on this family of boards are militant about the precise rules, and rewarded through gamification for doing so, thereby creating a hugely successful and useful franchise. So - the last point is just an observation, not a defense. The first point is my defense. – Jim Aug 18 '17 at 19:52

The plates can be removed, cleaned up and installed into another (new) hard drive. Cylinders that are fully outside the hole area should be readable no problem. This means, majority of the content, if there is only a single small hole.

The plates may be re-balanced by drilling another hole of the same diameter in the opposite side. Some means must be taken to prevent heads running over the holes, but looks possible.

There are two points here - what works, and what you should (or should not) do.

When I'm done with an old HDD, I open the top and heat it red-hot internally with a small DIY gas torch. It takes a few seconds from start to end. No magnetic data is going to survive the heat rise, which destroys/randomises the magnetic domains with absolute certainty, even if the plating on the platters wasn't oxidised/charred/burnt off and peeling. The case is easy to open too.

Notice the emphasis above: it's what I do. Almost certainly it isn't what you should do as a business. Nor is drilling, acid, electrocution, thermite, or any other fun activity. There are serious issues to consider before letting staff loose on the disks.

As an individual I'm fine doing what I prefer. As an employer your company is probably legally liable for staff safety and any accidents (in most if not all countries). I wouldn't allow my staff to do what I do personally. All it takes is one accident with a drill, due to exuberance or carelessness, some metal swarf to hit an eye, or anything else, and you can expect a visit from the lawyers who will ask you exactly what training and control your company gives, when it turns ugly.

Most of the alternatives suggested in other answers are a lot of fun - until they go wrong. At which point one person is in the line of fire. You.

Alternatives - top off case (ensures exposure as other answers state), and ideally some action that physically damages the platters (in any manner) but doesn't incite reckless conduct or risk an accident. Perhaps buy a hand-held demagnetiser (mains powered, produces a powerful local magnetic field designed to randomise data, has little or no harmful potential). Less exciting but a lot safer.

  • 1
    Our toolmaker apprentices are under supervision and using specialized drilling machines and wear protective gear. I think they'll be fine. – RubbelDieKatz Aug 24 '17 at 5:40

Considering that you''ve many drives to wipe, you may invest some time in an auto-eraser PC : barely a linux host that wipes any attached drive(s) and fill them with random data.

Then, destroy them physically :

  • For HDDs: an hydraulic press with a sharp end
  • For SSDs: 300V AC on the chips' pins.

Adjunct: In case you want to orderly disassemble a hard drive to dispose of the platters, there is a trick to know. If you attempt to remove all the screws on the plate stack one after another, the last one or two will always appear to be immovably stuck, and the torque you can bring to bear on it safely is severely limited by the screw still being in an easily rotated part. This is because the unequal tension on the top washer invariably jams it into the threads. Loosen all the screws evenly but only very slightly at first, only then remove them all the way.

Determined attackers will still be able to retrieve partial data and there are places I've heard that specialize in this sort of thing.

If you really want to wipe the drive's data, simply employ the Gutmann method (https://en.wikipedia.org/wiki/Gutmann_method). Then run the drive under a powerful electromagnet.

Anything short of this and you will at least have partial data recovery as possibility. That being said, unless the NSA or some organization with very high technical resources/skills is after your data, it's usually safe to just do a DoD 5220.22-M wipe(3 Passes).

The alternative is to melt down the drives, in which case data recovery is infeasible in any case.

Now the matter is different for SSDs, for normal attack vectors, a "secure erase" (typically available in your SSDs management software) is usually enough. TRIM does not erase data on an SSD, simply marks it as empty and ready for re-usage.

Otherwise, melting is still a good option :)

  • 3
    As the OP makes clear, the Gutmann method has been superseded by more up-to-date data. – MadHatter Aug 17 '17 at 5:38
  • 3
    I disagree with "Anything short of [Gutmann + magnet] and you will at least have partial data recovery as possibility." I have yet to see a single claim to be able to reliably recover data overwritten once with zeroes. That an attacker being able to do 6% better than random guessing via using an electron microscope means you don't have to worry about it. A single zero pass is sufficient to protect your data from anyone who isn't a state-level adversary... and even they are unlikely to be able to recover usable information. – TemporalWolf Aug 18 '17 at 17:31
  • 3
    Why do people continue to recommend a 20 year old technique designed to work with hard drives that haven't been manufactured for years and had a capacity much smaller than a cheap flash drive today? Even Gutmann says the Gutmann method is unnecessary, and he's been saying that for over a decade. – barbecue Aug 18 '17 at 18:47
  • 3
    @Akumaburn Gutmann even states right in the paper on his web site that "a few passes of random scrubbing is the best you can do". cs.auckland.ac.nz/~pgut001/pubs/secure_del.html If you're going to try to go against the person who studied how to reliably overwrite data on MFM and RLL media on the difficulty of recovery on modern media (which basically means anything made within the last two decades or so), you'll have to do better than unsubstantiated claims. – a CVn Aug 21 '17 at 15:18
  • 4
    Formatting a drive != a zero pass. Neither of those will recover anything from a single-pass zero wiped drive. From easeus: "Formatting a disk does not erase the data on the disk, only the data on the address tables." – TemporalWolf Aug 21 '17 at 18:09

Well seeing as Zero-ing out the disk using military grade writing doesn't suit your fancy, a sledge hammer will do the trick better than a car or drill press. my preferred method: metal CNC (Fusion 360 or find an online 3D model of a hard drive and feed that into a linux CNC) and watch the machine battle the machine! Gloriously calculated "paths" for the mill to follow!

Note that the military has their hard drives embedded with thermite (no joke) and they just shoot the darn things till they melt. Ooo nifty idea, take a torch to them, or cook them in an aluminum oven!

Personally the magnets are awesome and the platters make good mirrors so my bad drives are on the wall :D

  • 3
    What is this "military grade writing" of which you speak? – a CVn Aug 17 '17 at 10:53
  • 4
    @MichaelKjörling capital letters, I'm guessing? – MadHatter Aug 17 '17 at 15:18
  • 2
    Have a link confirming that the military (US?) actually puts thermite around their hard drives? That sounds insanely dangerous, computers (especially laptops & batteries) do overheat & catch fire sometime, it doesn't strike me as a safe practice. And where would they find the space for thermite anyway, laptops don't have much spare room, or are there really HD manufacturers that would stock thermite & put a sliver of it inside a HD, safely in their factory? – Xen2050 Aug 18 '17 at 15:57
  • 8
    @Xen2050 “Note that the military has their hard drives embedded with thermite (no joke) and they just shoot the darn things till they melt.” Yes, the original poster should please provide some context to this fairly ridiculous claim. It’s the kind of claim some self deluded “mall cop” would make. – JakeGould Aug 20 '17 at 16:28
  • 5
    @Tmanok So it's really only second-hand info from one friend of yours? Others in the military dispute the claim (Steve). Not that I don't believe you, I'm sure your one friend could've said that, but without any proof you should question your friend's claim too. (Just an idea, but for example was there any alcohol around when he made the claim?) – Xen2050 Aug 21 '17 at 6:18

DBAN may be preferable up to a given size. If you have to wipe disks containing terabytes of data it will take hours to wipe it completely.

Using a drill takes only a second. I'll let you do the math yourself what makes more sense for you.

  • 3
    But what's your answer to the question? You seem to state that drilling a hole makes all the data unrecoverable, is that correct? – pipe Aug 16 '17 at 13:44
  • 2
    It makes it unrecoverable for the vast majority of use cases. As the later answers explained in detail, it depends on the threat model if it (while it is the fastest method) is sufficient enough. – Gerald Schneider Aug 16 '17 at 13:54

Depends on the requirements of your enterprise and also the level of the group attempting to reconstruct your data.

Casual hackers may be defeated by this but some specialized agencies can get past, about, 6 layers of "wipe" and can also recover all the data but the hole (On raid though, they can usually reconstruct that, depending on whether they have some of the parity drive etc.).

In our copany, we bulk demagnetize and shred most. Some projects also require that the drives are disassembled and the individual disks sanded,.. after the demag and then a shred.

  • 10
    "our agency can get past, about, 6 layers of "wipe"" is a pretty tall claim with no supporting evidence. I suspect there are a lot of caveats to that statement. – TemporalWolf Aug 16 '17 at 18:00

protected by Sven Aug 17 '17 at 12:17

Thank you for your interest in this question. Because it has attracted low-quality or spam answers that had to be removed, posting an answer now requires 10 reputation on this site (the association bonus does not count).

Would you like to answer one of these unanswered questions instead?

Not the answer you're looking for? Browse other questions tagged or ask your own question.