Stuff piles up. You know how it goes. Perhaps the worst of it is down in my workshop, where I’ve done nothing ambitious in almost a year. (My steampunk Geiger counter was the sole exception.) Predictably, when something comes to hand and no good place for it is found in a few seconds, it goes downstairs and ends up on my workshop floor. I still have some work to do down there to make the place habitable again, which brings me to the question of degunking hard drives.
What I’m referring to here is the pile of old hard drives on my workshop shelves. They’ve been there a long time, and they’re taking up space. I took a look at them earlier today. One of them is an 80GB drive of 2003 vintage that I took out of my old Dell Dimension before scrapping it last year, and that’s worth keeping. Most of the rest of them date back to 1998 or before. Many are not even mine. People I barely know have given me mid-90s vintage PCs, from which I generally pull the drive and SIMMs, and then take to the local computer recycler. What I haven’t done in some time is look at the capacities of the drives. That was an eye-opener: Two were 1.2 GB; another 3.5 GB, two more were 4 GB, another 6 GB. The biggest was an 18.3 GB Seagate Barracude, which may sound useful except that it uses the “wide” SCSI interface common in high-performance desktops in the late 1990s, now present on no machine in my collection.
They have to go. I used to dismantle old hard drives to pull the magnets out of them, but I already have a bin full of hard-drive magnets. I suppose I could connect each of them in turn to one of my machines, run Eraser on them a time or two, and then give them to the recycler. (After all, some of the data on those drives isn’t mine, and I no longer remember which drives are which.) Or…
…I could use One-Clunk Degunking: You place the drive in question on the driveway, and give it one good hard clunk with a five-pound sledgehammer. (I have granite boulders in abundance, which spares my concrete in case I miss.) That should do it, but as the incremental cost of clunks is small, two or three more for good measure won’t hurt. (Won’t hurt me, at least.)
I remember when hard drives (and the computers they were in) cost a great deal of money, and it’s tough not to look as those drives and think that whacking them is a terrible waste, but no other uses come to mind, and my shelves are pretty full. So out I went a few minutes ago, laid six drives on a flattish boulder, and gave each one a good hard clunk with the sledge. I was a little disappointed that they didn’t look more destroyed than they did, but trust me: No one will be reading those drives again. At this point (with plenty of shoveling still to do down there) I’m good with that.
“Students find U.S. defense contracts in Ghana on $40 hard drive ”
http://www.geek.com/articles/chips/students-find-us-defense-contracts-in-ghana-on-40-hard-drive-20090625/
An old friend worked on a highly classified US Goverment project that came to an end several years ago (late 80s or early 90s), and he told me with some glee about the infosec procedures they were given for the hard drives:
1. Format
2. Format
3. Format
4. Degauss
5. Sledge
He reported that his team took turns taking their frustrations out with step 5.
Back in the day, I was in an Army unit with lots of cryptographic equipment. If we ever had to destroy that stuff, it would have to be done quickly and thoroughly. There would be no time for a five-step process. We had a one-step process in readiness:
1. Thermite
Of course, for civilian use, thermite is had to get, illegal, and unsafe. It would be cool, though. đ
I didn’t know thermite was illegal. Also, from what I’ve read, it was not quick or easy to get going–though, of course, once it got going there was very little you could do to stop it.
Most of what I know about it comes from my readings on railroad tech, where it was used to weld rails and other low-precision iron things out in the field. I think Lindsay has a whole book on it, though I’m sure that book is now 70 years old. Arc welding and high-tech gas welding pretty much put thermite into the shadows.
I don’t actually know whether thermite is legal or not — I just guessed based on what all else is currently illegal.
Yeah, it is hard to light, but it is very thorough. For our purpose it was important that even a knowledgeable person be unable to tell even the kind of thing we had destroyed. With a disk drive you usually just need to insure that it is truly unreadable; it’s OK is it still resembles a (battered and broken) disk drive.
Thermite isn’t illegal at all, or even restricted.
At each of the last few Berzerkers, tons of thermite have been burned off.
One year, a thermite birthday cake. Another year, a thermite volcano with a fully charged propane tank buried in it. (There was no explosion, just a huge gout of flame.)
And yes, thermite does a really good job of “erasing” a hard drive.
In the above five-step method reformatting the drive is mentioned. I think it is useful to point out that reformatting a hard drive does not erase the user accessible data on the drive (this is also the case when deleting files); it just overwrites the file pointers. The data should all still be intact and recoverable until the user overwrites it with something else.
With the one-clunk method, in theory the data should all still be there even though you have effectively made it very difficult to access. With sufficient funds, someone could potentially still use an exotic method, such as putting the platters in a new drive enclosure (after repairing the platters if they are bent or cracked) or even examine the platter surface with a technique such as electron microscopy, to extract the data.
The standard method of repeated overwrites of random data should eliminate the possibility of anyone ever getting the data, but on large drives this can take a prohibitively long time. Things like thermite or explosives could work well, as mentioned in the previous comments. However, unless the drive is destroyed completely I would be skeptical that the data is unrecoverable (it has been demonstrated that hard drive data can be recovered even after exposure to extreme heat and combustion contaminates: http://www.pcmag.com/article2/0,2817,1911114,00.asp).
Here is an alternate technique I thought of a couple of years ago: disassemble the drive and then grind off the magnetic thin film on the platters, either with sand paper or by some other method. Keep grinding until you get all of the film off.
Well, sure. If I were trying to hide nuclear secrets or something, I would do a much more thorough job. The drives pictured contained old copies of Windows, installed apps, a lot of boring email, some CD rips, and other odds and ends of no great significance.
If I wanted to make them even harder to recover, I’d toss them in with the weekly haul of doggie poop bags, making sure there were no identifying scraps of paper in the same bag. Once they’re in the landfill, surrounded by tens of thousands of similar black garbage bags full of rotting leftovers, dirty diapers, and dog poop, I doubt they would be findable, and if you can’t find them, you can’t recover them. I’m guessing that there are a lot of hard drives under 3,000 feet of water off the Atlantic Coast. I doubt that anybody even bothered to whack them.
Is it really true that formatting a hard drive does not erase the user data? You are correct that doing an ordinary delete of a file merely marks it as not there, without erasing any of the data. And what Windows calls a quick format just initializes the directory, free space list, and similar things without erasing the user data.
But I am under the impression that a regular, not quick, format, at least in Windows, erases every sector of the drive. It seems to take long enough to be writing every sector. I certainly could have the wrong impression, and if so, I would like to be corrected. Can you refer me to some more or less technical description of what a format does on reasonably modern hard drives?
I am aware that even a one-pass erase or overwriting of disk sectors does not make the data unrecoverable for someone with advanced tools, so I’m not especially advocating it as a method for disposing of sensitive data. I am merely curious about whether a normal format operation overwrites all of the sectors on the disk or not.
Good questions: I know that a âquickâ format doesnât destroy user data but why doesnât the ânormalâ format erase user data, and if it doesnât then what is it about the ânormalâ format that takes so long?
Disclaimer-
Donât apply any of the following information to flash based disks; they are a different situation entirely.
Short answer-
Similarly to the âquickâ format, the ânormalâ format only rewrites the table of file pointers at the beginning of the disk (the files themselves are in general left untouched). The reason the ânormalâ format takes so much longer than a âquickâ format is that in addition to overwriting the file pointers it also checks the disk surface for bad sectors and remaps any it finds (during this process the files are again left untouched, although some of the file containing sectors may get moved if they were found to be bad).
Less short answer-
I think a major source of confusion is the mixing of the formatting and zero fill concepts, when in fact those are two distinct processes. Here is a list of some basic hard drive operations: low-level format, ânormalâ high-level format, âquickâ high-level format, ânormalâ zero fill and âquickâ zero fill. Notice that there is only one type of low-level format, itâs essentially the last step of the disk manufacturing process and not something that end-users normally do to hard disks (http://www.pcguide.com/ref/hdd/geom/formatLow-c.html). The ânormalâ high-level format rewrites the file pointer table and does whatever tasks are needed to prepare the file system (NTFS for example) and get the drive ready to accept user data, this in general leaves alone the part of the drive that actually contains user data. A âquickâ high-level format is the same as a ânormal formatâ except that it also scans the disk for errors and remaps any bad sectors that are found. âNormalâ zero fill simply writes zeros to the entire disk surface (as pointed out, the data would probably still be recoverable since zeros arenât random). âQuickâ zero fill writes zeros to the first couple of sectors of the disk, saving time while achieving basically the same end result as a ânormalâ zero fill. The zero fill is sometimes referred to as âreinitializationâ, since it basically returns the drive to its ânewâ state.
Sources-
I wish I knew of a book, document or webpage that explains disk formatting in a clear, complete and concise way. All I really know to do is direct you to Wikipediaâs article on disk formatting, which cites several useful sources of information and is fairly complete (but not outrageously clear or conciseâŚ): http://en.wikipedia.org/wiki/Disk_formatting
In particular, the above Wikipedia article cites The PC Guide (http://www.pcguide.com/ref/hdd/geom/format.htm), which contains the most detailed and technical hard drive formatting information I know of. If anybody reading this knows of a better source of technical information on disk formatting please point it out. Also, there may be some formatting related information hiding somewhere in SpinRiteâs documentation (http://www.grc.com/srdocs.htm)
I love the number of responses to this post; The sledge hammer is a great idea. Last time I wanted to trash a drive, I found the drill motor in the garage before the sledge hammer. A couple of 3/8″ holes fixed it right up.
One reason I think the sledge is better is because it damages almost everything in the drive: The platters, the spindle, the PC board, the case, everything. Drilling holes still allows the drive to be dismantled, and (unless you’re willing to wear out a bit or two in the drilling) may leave enough unmolested oxide for snoops to get something. Bend or crush everything, and they would have to work a great deal harder.
Besides, the job is easier on the sledge than it would be on the drill bits.
The last hard drive I took apart completely was in 2006 or 2007 and at that time the platters were made of GLASS not metal. They looked like metal, but when wrapped in a few layers of newspaper and whacked with a mallet there was nothing but silvery looking powdered glass. Now this was at work and the drive was from a server and if I remember correctly it was probably SCSI from a RAID array so that may not be the norm, but I can’t think of anything more secure than a pile of glass powder that you could commit to the four winds.
For drives that work in a box going out of service a few passes with one of the options on DBAN is good enough for anything I would have on a drive.
Oh, back in the early 1970’s in the Air Force I was detailed once to run some very classified mainframe tapes through a very powerful degausser twice — and then had to strip the thousands of feet of tape into a burn bag to be burned. I never guessed that I would be dealing with these same issue decades latter!
I use the same method for destruction of hard drives, but I take the cover off the drive first, then apply the sledgehammer to the platters directly, a number of times.
Of course, once I made the mistake of doing this to an IBM DeskStar (now made by Hitachi) drive, on the kitchen floor. The platters in those drives are not made of metal, but glass. CRASH! Destruction had to be halted while a billion little fragments of glass got swept up. Since then, I do all my destruction outside, on the asphalt behind the dumpster.
Smashing hard drives with a sledge hammer and blowing the dust away from a PC chassis are outdoor sports.
“Cleaning dust from inside computer”
http://www.youtube.com/watch?v=58OM5ZTznTE&feature=related
Absolutely. As other commenters have said, some older hard drives had glass platters. I’ve never opened up one of those but I’ve certainly heard about them. As for dust, I periodically take my main machine out to the garage, but a ground clip on it, and then blow my shop vac at it. If I do it often enough it doesn’t generate a dust storm, and it’s definitely an issue in keeping the system cool internally.
A 30’06 will do the job. If you want a nice challenge, try it from 200 yards. It costs more but it’s also more satisfying.