Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 


Forgot your password?
Close
typodupeerror
×
It's funny. Laugh.

The Pentagon Discovers dd 264

Posted by CmdrTaco from the god-bless-america dept.
hph writes: "CNN.com reports that 'The Pentagon believes it has found a way to give its old computers away to American schools and still protect information locked in the machines' hard drives.'" I hope this story amuses you all as much as it did me.
This discussion has been archived. No new comments can be posted.

The Pentagon Discovers dd More

The Pentagon Discovers dd

Comments Filter:

  • Re:Whats dd (Score:1)

    by Anonymous Coward
    For you non-UNIX people here at Slashdot (both of you), 'dd' is a command that reads and writes raw bytes to files or devices. The usage that the article is implying is "dd if=/dev/zero of=/dev/hda" which will overwrite the primary hard drive with zeros, destorying all data. (Well, obscuring it to the point where only trained technicians could possibly recover it by magnetic analysis of the physical hard drive platters.)

  • Bit more complex (Score:1)

    by Anonymous Coward
    It's a bit more complex than dd. Using high-tech microscopes, etc. the NSA can uncover data that was overwritten. Some tools overwrite the data multiple times, making rediscovery much more difficult. The story becomes even more complex when you introduce modern hard drives that can move data on the platter around at will to hide partially bad blocks on the drive. You can still have recoverable classified data on those bad blocks. Designing tools to properly wipe data from drives is a non-trivial task.

    - pmitros

  • Re:In other news... (Score:1)

    by Anonymous Coward
    Today it was discovered in a classroom in Denver, Colorado that when discarded chalkboards were sprayed with a phosphorescent acid group, previously erased markings became visible once again. The Department of Defense is trying to recall its previously donated chalkboards, but a militia on the outskirts of Denver has acquired twenty such boards and is defending them with a number of recently built suitcase nuclear weapons and H132 military grade sniper rifles that were under development at the Pentagon weeks before.
  • Simple, just peel off the top 20 or 30 layers that you wrote and you'll see the lower layer easily. Believe me, we do this in forensics. The best way to wipe your drive is with a sledgehammer and then burn the rest of it in an incinerator.

  • Serious Issue (Score:1)

    by Anonymous Coward
    I've worked with a non-profit group partnered with the US Navy to refurbish computers for schools. Letting sensitive data out is a serious problem. Non-classified stuff being left on the drives is a frequent issue. You name the unclassified materials, we've seen them. The real problem is the actual classified stuffwe run into sometimes. It may be 1 in 1000 computers that have something like this on them, but sometimes people are really stupid. You would be amazed at the kind stuff people let out. We delete it all and reformat the drives, but it bothers me that no one checked to make sure they didn't leave stuff on it. For awhile the Navy started using some sort of microwave to clear the drives, but IMO (since none of them worked afterwards) this just fucked up the drives making them useless for schools as well. Thus, I think they should just destroy anything that EVER held classified data. Burn, crush, shred, melt, bury it, or throw it into the Atlantic rather than take a chance. At the same time, I no longer wonder how the chinese stole nuclear secrets from Los Alamos given the piss poor data security procedures I've seen here.

  • Re:Wow (Score:1)

    by Anonymous Coward

    is that really you? you're damn hot...

  • Re:Its only UNCLASSIFIED data... (Score:3)

    by Anonymous Coward on Sunday June 10, 2001 @11:23AM (#162261)
    There is a principle in DoD security, that the pattern of usage of Unclassified data should be classified, because it can give an indicator to other, more confidential information. Information Analysis has always been the spook's best friend, because people don't think to watch out for the 'trivial' stuff.

    For those of you who remember history, see also the US's exploits in the Pacific during WW2, accurately outguessing the Japanese, based on whether they would transmit after they leaked information to them. There are many other instances where this has been helpful.

    Just because there are occasional individual misses in a large organization, do not make the mistake of assuming that the overall practices of that organization are lacking rigor. Human error is a constant problem in every outfit.

    As for Slashdot's snide little comment about DD, just writing a single null bit is most definately NOT up to DoD standards for deletion of sensitive data. Magnetic media has a tendency to maintain shadows of earlier data which, using sufficiently sensitive apparatus and diligent study, can yield a surprising amount of information that could be considered 'deleted'.

    Personally, I would advocate a limited-lifespan design where two drives are maintained. One with a pad of entropic noise, one with the actual data, encrypted with this pad. As a sector is accessed on one, it is decrypted or encrypted using the noise. At the end-of-life, the pad drive is pulverized, /then/ a traditional 7-write delete could be used to wipe the data.

    But that would require foresight, and that sort of thing would never make it past congressional accountants.

  • dd was too slow for us (Score:3)

    by Anonymous Coward on Sunday June 10, 2001 @11:50AM (#162262)
    When I was in the service, when we had to get rid of a hard drive, we would use it as a test machine during "bug out" drills.

    Our normal systems had classified (conf, secret, ts, ts+codeword) info on them. During wartime, if we were over-run, we would set off incendiary grenades (thermite, for those of you playing along at home). Since it is a rather startling sight and you want to MAKE SURE that it is done correctly, we would have one live drill per year.

    Alternatively, we would put the TS ones through the shredder. We would take apart the drives and then feed the platters through. Came out as powder. Needless to say, this was a "no-tie" area.
    I did have one captain that made us put the head through as well, since "the field could store some information." Yeah. 1 bit...
  • Because restoring the data tends to rely on rather expensive equipment. A drive with double the capacity, and 10000 times the price wouldn't sell that well.
  • There is a clearance level called sensitive, and confidential. All files must be marked properly if they contain this information. In which case they are not unclassified. As such, records must still be kept of how many copies and where they are located if memory serves me correct.

    The first problem with that is that from a strictly operational viewpoint, you can't classify *everything* -- because at that point you've eroded the value of the lowest security classification. If every document that is generated is marked sensitive, then either everyone will have to treat every book, file, floppy disk and piece of paper as sensitive ... or everyone will ignore the sensitive classification and keep working as usual. Human nature.

    The second problem is the one that I raised, and that you ignored -- individual pieces of information by themselves are often completely benign, but correctly correlated can produce a very accurate estimate of information that would normally garner a much higher security classification.

    This applies in every day life, as well. As an example, take the current trend in grocery marketing -- the saver club. If you join the club and use your club card everytime you go shopping at a particular grocery store, you get price discounts and cash back after enough purchases. Of course, the reason the store is willing to give you those discounts is that they are gradually building a very accurate database of information about your buying habits: what brands of food, how much, how often. You might not care if someone knows you eat Cheerios, but how about condom usage, or specific medicines? There's a lot of personal information to be gathered by aggregating supposedly non-sensitive data...


    Are you moderating this down because you disagree with it,

  • Re:Its only UNCLASSIFIED data... (Score:5)

    by miniver ( 1839 ) on Sunday June 10, 2001 @11:04AM (#162265) Homepage

    Congratulations. You have failed the first test for a security clearance. Just because the data is unclassified, you can't assume that the data isn't still sensitive.

    As an example, suppose you have a spreadsheet that details the fuel consumption for a group of F-16s, and another spreadsheet that details fuel purchases for an air base. Individually the spreadsheets are unclassified information, but together they are sensitive, because from the two spreadsheets you can deduce a great deal about the missions and deployments for that group of F-16s. Even months after the fact, that information is still sensitive, because it can be used to make predictions about that air group.


    Are you moderating this down because you disagree with it,
  • Not entirely true. Personal data (SSAN, DOB, etc) is considered "FOUO", or "For Official Use Only". This marking is actually a 'security handling' marking rather than an actual classification. (Not sure what you did in the Army, but consider a document that is SECRET// - it's actually SECRET, but can only be handled in the ways appropriate to that program.
    And there is a TON of useful information that can be gleaned from UNCLASSIFIED data, as mentioned in numerous other posts here. Personally, I think that all drives (CLASS and UNCLASS should be destroyed. Media is cheap enough these days that a school could buy a smallish but usable drive for about 100 bux and have the rest of the hardware for free.
  • clean-room zapping, and re-assembly would be better, but what about the cost? is it worth x $100 drives?
    (not a flame, just an honest question)

  • Re:Sorry that is *NOT* DOD standard. (Score:4)

    by psychosis ( 2579 ) on Sunday June 10, 2001 @02:01PM (#162268)
    Just a few minor points:
    SCIF = Sensitive Compartmentalized Information Facility
    You are pretty dead on about the requirements to certify a facility though. (I worked with certifying them a year or so ago.)
    Within the past few years, the tide has changed a few times on what to do with media. Currently (to mean that this article is not legal guidance in my mind), the standard is to physically destroy and burn all media regardless of classification. My guys have a locker full of around 50-100 drives that are waiting to be sledgehammered and sent to a burn facility. (Not sure if this is a "lots of fire" one, "acid bath" one, or something I'm not even aware of...)
    And overwriting ANY number of times is not enough to save it from the tools that a foreign intelligence agency would have. I'd assume that we have similar technology, but that's some of the cool stuff I'm not privvy to. ;)
    I'll be interested to see what guidance might be waiting for us when we get to work Monday... And yes, those platter clocks/plaques do kick ass!
  • I'll just point out that you're attributing an unwarranted level of competence to the various "Intelligence" communities.

  • dd would only perform a single write, which is in no way enough to delete all traces. That would be like erasing an audio tape; tiny vestiges of the original recording would still be present.

    Uhh, no shit Einstein! And if you run dd twice it would perform two writes. What a concept! Amazingly, if you run dd 10 times, it will perform 10 writes. Wow!
    ___

  • It' probably has do do with economics.
    I may have a similar situation. I work for a hospital, and we want to get rid of a stack of 486's sitting in a storeroom. Some are chock full of patient info. HIPPA (patient privacy act) says $10,000 fine for each incident of improper release of information, so we have an incentive. Not to mention how happy attorneys are to discover such incidents. (It can get real expensive)

    First you have to plug the thing in and hook up the cables. There's only table space for two at a time.
    Second, the overwrite. I'm only doing one overwrite. Why?
    Do you have any idea how long it takes to do a single overwite of a hard drive on a 486, much less 7, 10, or 16 overwrites?
    Suppose the 486 is flaky and the floppy is broken.
    I'm supposed to fix it so I can erase the HD?
    The salary cost of personnel doing thorough overwrite, degaussing, and all those other games makes it a hellavu lot cheaper to incinerate the drives and buy new ones. (if one were to attempt a thorough job). How many is the Gov talking about? 10's of thousands?

    Ideally, one would have them wiped at the desk before installing the upgrade. (as if I trust the contractor-of-the-day to remember or actually spend the time).

    Reading between the lines of the article, and interpolating from my own experience, the real problem with releasing overwritten drives versus the acid bath, is that some of the ones that were supposed to be overwritten were not done. How would you know (cost-effectively) that the overwriting was done completely, or at all?
    Double-checking doubles the cost.

    What's the point in giving away a $100 dollar computer if it costs $200 in taxes to clean it up?

    Well, the schools don't care if it costs the government $2000 to donate a 486. To them, it's "free". People holler, politics intervenes, and we're back to giving them away.

    The Pentagon says "to hell with it", we'll just do a single wipe and get rid of the problem.
  • Since the drive reads data in a track according to a somehwhat fuzzy technique that attempts to read most of the data most of the time from the area of the disk it expects to be able to read from, it is also true that data gets written only to approximate locations in a track. That is, there is bleed over into synch tracks and other areas which aren't typically overwritten by software commands. There are also redundant tracks that are used to compare read results so wiping out one logical track needs to be reflected in all logical tracks as well.

    Otherwise they have to physically destroy the drives including crushing and burning.

  • How I learned to stop worrying.. (Score:3)

    by Splat ( 9175 ) on Sunday June 10, 2001 @03:47PM (#162281)
    I work at a school and over the past year we've received over 10 donated 486 laptops from the Nuclear Regulatory Committee whose hard drives weren't wiped in any form whatsoever.

    Mind you, I haven't found g any data on them, but they still had an operating system + programs. I was really shocked when I booted them up and their DOS Batch menus popped up "NUCLEAR REGULATORY COMMITEE .... 1 Wordperfect ..... 2 Windows" etc. Funny, but disturbing. If anyone did "cleanse" these computers before they donated them it was simply by using "DEL *.DOC". Yeah - real secure...
  • This strikes close to home, since I am in the process of excessing a large quantity of old and broken computer equipment. While I would like to sanitize all of the equipment, the problem is that much of it consists of non-functional computers and old hard drives. It would be more cost effective to destroy all of the hard drives than it would be to try to erase them. I'm not sure if the property management rules take that into consideration.
  • If a new hard drive costs "a couple hundred bucks" you are buying them in the wrong place. You can easily get a new drive for under $100 if you know where to look. Of course we are talking about the government, so they probably pay way too much, but still... They could just remove the drives and ship the schools the machines and the schools could buy the new drives themselves. Still cheaper than buying a whole new computer.

  • dd is often not enough (Score:3)

    by Adam J. Richter ( 17693 ) on Sunday June 10, 2001 @11:32AM (#162288)

    With specialized equipment, you can often read data that was overwritten on a hard disk, so dd is often not enough. See Secure Deletion of Data from Magnetic and Solid-State Memory [sourceforge.net], by Peter Gutmann of the University of Auckland. There is also a previous slashdot article [slashdot.org] on this subject.

    Although the article says that hard disks that held classified data will still be destroyed, there still is lots of information that is unclassified but is not supposed to be released, for example, private personal information, a large body of information that should have been classified because classified information can be derived from directly from it, etc. You can bet that the US military is much more familiar with this issue than the insulting "Pentagon Discovers dd" slashdot title implies.

  • Re:Its only UNCLASSIFIED data... (Score:3)

    by dillon_rinker ( 17944 ) on Sunday June 10, 2001 @11:02AM (#162289) Homepage
    It's been a few years since I was in the Army, but IIRC, personal data is classified 'Confidential'. Or at least it's treated the same as confidential-classified material.
  • they could get it back... the boys at Fort Meade specialize in recovering third- and fourth-generation overwritten files.... my brother-in-law does such things for a living (though not for Never Say Anything)...

    Of course, I noted near the end of the article that the truly classified machines still get trashed, it's just the garden-variety receptionist and lower-end lackey machines that get given to universities where the 33l33+ #@X0R d00dZ lurk...

  • Hmmm! Sounds like it would definitely be cheaper for them to stuff a new drive into the box and *then* give it away to the school!


    "Give the anarchist a cigarette"
  • Erasing data is harder than you think. Even dd if=/dev/urandom of=/dev/hda probably doesn't erase all data beyond recovery. For maximum effect you need to do several rounds of erasing with certain bit patterns designed to maximise the effect on the magnetic patterns on the storage media as well as several rounds of cryptographically secure random numbers.

    Securly erasing magnetic media beyond any hope of recovery without destroying the media is *hard*.

    For a full account of the problems involved, read this [rootprompt.org].

    -henrik

  • Re:not very interesting (Score:5)

    by RovingSlug ( 26517 ) on Sunday June 10, 2001 @11:17AM (#162299)

    Agreed. Slashdot's presentation totally misrepresented the actual story.

    See, rather than applauding the Pentagon for giving away (!) computers (!!) to schools (!!!), and rather than commending the Pentagon for reversing an existing policy (the path of least resistance would have just destroyed those hard drives), Slashdot decided to flex its techno-elitism and show just how snobby some geeks can be.

    So, if some people at Slashdot would stop desperately trying to mock any and all mainstream journalism about computers, perhaps they'd see the actual value of this story.

  • (May as well reply to the replies to my reply here. Most of the arguments are in a similar vein)

    I stand my ground that 20 year old secrets are quite lame and not worth protecting. Your post was the one that seemed most rational, as it focused on things such as spy networks. No problem. That almost makes sense. If the network hasn't already been compromised (Aldrigde Ames, et al.)

    If foreign powers (and in this case, I think we need to primarily concern ourselves with the Soviets and possibly the Chinese) are incapable of breaking the secrets after 20 years, they aren't a threat. If they are capable, destroying the drives is a moot point; they already have the information. But I will grant that of all the arguments, the question of spys makes more sense than any other.

    Second point that many others made was WRT technological advances. Which doesn't float at all. Even after having a U2, I don't remember tales of a Soviet counterpart (I'm not a hardware buff, so I could be wrong. Still, it would be an important part to the Gary Powers legacy if there was evidence of the Soviets reverse-engineering the thing). There is also the more important matter of build-quality. You can have the greatest design in the world, but if your metallurgy/construction/operation of a device is faulty, who cares? I think the SCUD's proved that point about ten years ago. While the patriot missile helped, so did the fact that the SCUDs were put together like a Trabant. (Come to think of it, a Trabant in a trebuchet would likely have been more effective).

    (Most of the rest deals with the other replies, so don't take it personally if it's not 100% related to your post)

    Another poster mentions chemical and biological warfare, as well as the Manhattan project, as being items that are still rightfully under wraps. Give me a break. Without access to that data, China, Pakistan, and India (among others) all have nuclear programs. Concurrent discovery of technology is the norm, and the US didn't do anything grand, except get it done before having to get on with the island hopping.

    Biological is perhaps the biggest joke there is. Anybody with a few credits of chem or bio in college could develop a rather nasty thing to rain all sorts of shit down on an enemy. The real trick is delivery systems. Given that little GPS powered 'RC' plane, I don't think we need to go high tech, either.

    But of course, someone could steal the super-duper-top-secret GPS error removing protocol. This and the exact capabilities fall into the same category: big freakin' deal. Close only counts in horseshoes, handgrenades, and nukes. Do you think Saddam or Osama cares if their pathogen hits at 1602 Penn. Ave. instead of 1600?

    I could go on and on about why the arguments posted up to my post were wrong, and did not support the destruction of hard drives. BUT, I have largely reversed my opinion, based on one small thing that I haven't (yet) seen mentioned: most of these machines don't have 20-50 year old data on them. Most like, it is just a few months. Thanks to the DOD (and their worldwide counterparts) Intel, Western Digital, and the rest continue to make 386's, one GB drives, and 30 pin SIMMs. Those 486's that could go into the schools don't contain ancient information. They contain the latest and greatest, given the slow speed of replacement of computers by the DOD.

    (And to the moderator of my original post: If you think I am a troll simply because of strong language or an opinion that differs from yours, say so. Don't hide behind the 'overrated' tag. Obviously, at least four people on /. felt it was worth replying to with reasonably well thought out arguments (even though I disagreed with 99% of what they said). If you picked 'overrated' because you weren't sure if you were right, than you shouldn't be modding. And if you did it to save your karma, that's right, you are a karma whore. Earn your karma by posting.)
  • >>Is it just me, or does it seem to anyone else like Slashdot's editors 1) can't read, and/or 2) are easily amused?

    Since most of the readership falls into both of these categories, why shouldn't the editors?

  • Re: [accusation of gross incompetence] (Score:3)

    by Catullus ( 30857 ) on Sunday June 10, 2001 @03:05PM (#162308) Journal
    [accusation of plagiarism]

    [link to kuro5hin story [kuro5hin.org]]

    [patronising comment]

    --

  • Re:dd is not good enough to erase data (Score:3)

    by rabidMacBigot() ( 33310 ) on Sunday June 10, 2001 @11:35AM (#162311)
    True, binary digits are either one or zero. But binary digits on a hard disk are written into the real world, and in the real world, the one or zero is the magnetic ailgnment of a few particles of ferrous oxide. These particles are altered when they're overwritten, but there's a very good chance that particles in the space between tracks may keep their original alignment. The trick, then is to overwrite the disk with many random bits, so that the patterns that might be found on a disk full of zeroes get lost in the noise.
    At least, that's how I think it's done. Feel free to correct me if I'm wrong.

    --
  • I stand my ground that 20 year old secrets are quite lame and not worth protecting. Your post was the one that seemed most rational, as it focused on things such as spy networks. No problem. That almost makes sense. If the network hasn't already been compromised (Aldrigde Ames, et al.)

    So, in other words, you're saying my point about spy networks would isn't valid, since the secret had been blown. There's a bit of a problem, though: If there is any 20 year-old info that needs to stay secret (and has), how the hell am I supposed to know about it to rebut your argument? The government probably has all sorts of stuff that's 20 years old that is still secret, and it's not about to surface on Slashdot.

    I have largely reversed my opinion, based on one small thing that I haven't (yet) seen mentioned: most of these machines don't have 20-50 year old data on them. Most like, it is just a few months.

    I wasn't saying the data was necessarily old now, but that, even if the technology doesn't currently exist to recover the drives, it could still be a problem if such technology is developed in 10 years. If they're just retiring the computers now, then, in all likelyhood, the data isn't too old.
  • for i in `seq 1 16`; do cat /dev/urandom | cat > /dev/hda; done what the fuck can you recover after that? you can't do wizard things... What the hell can the NSA know more about filesystems that us, who understand and write filesystems?

    There is a world beyond software, fool. Read this article [auckland.ac.nz]. In case you're too lazy to click the link, you can recover data from a drive by disassembling it and using magnetic force microscopy with a scanning tunneling microscope. Even after you overwrite a bit, there are still traces of its previous value.
  • Interesting argument. And one that I am not sure I can refute.

    Touche.

    First of all, thank you for the civilized debate. It's a rare thing on slashdot... You seem to be a rational, open-minded individual, and the world needs more like you.

    I guess we've come to the conclusion that the government may have important 20 year-old secrets, but we can't reasonably find out if they do.

    I figure their security experts probably know best, so they should be the ones making the decisions, not politicians who think the internet is synonymous with the web.
  • All Soviet troop movements from the spring of 1979 to the summer of 1980,
    The latest blueprints for their newest fighter,

    These are fairly lame examples, but I'll work with them anyway... Troop movements would reveal tactics and doctrine, which don't change very quickly. Fighters and other military hardware often stay in service for 30 or 40 years (think--when was the F-15 introduced). Hell, the Russians are about 20 years behind the US in sub quieting technology, so getting old info there could probably still help them, and then we'd see the technology show up in subs sold to the Chinese (and, in turn, to every two-bit rogue nation in the world).

    A much better example of something that would still need to be secret after 20 years would be the names of agents operating in foreign countries. Admittedly, that would be classified, but the original poster was talking about releasing drives that had held classified data (and I pointed out that it's hard to be sure a computer wasn't ever used for classified stuff).

    Why take chances with national security just to get some crappy 486s into schools? For the cost of proper data wiping (remember, the Pentagon never does anything cheap), you could probably buy them Pentiums.

  • Politics and Security don't mix (Score:5)

    by tbo ( 35008 ) on Sunday June 10, 2001 @11:26AM (#162319) Journal
    This is a very bad decision, which I'm sure is being mad for political and not security reasons. From the article:
    Others supported it after an audit found sensitive information such as lists of names and addresses had been left on hard drives of donated computers. Though unclassified, they said such cases still present risks.
    This is entirely true. There's a lot of information that, when taken individually, isn't dangerous, but, when combined with large amounts of other info, could present a security hazard. Maybe one piece of unclassified info says the Air Force is building a new stealth fighter, and another piece of info is an Air Force requisition order for 20 tons of titanium. Put 2 and 2 together, and you know that the new fighter will be made of titanium... (Example borrowed from The Cuckoo's Egg [amazon.com]). Also, are you sure that none of those computers was ever used for classified information?

    Now I know the Slashdot editors (and more than a few Slashdotters) think that they're left-wing 31337 political h4X0rs, and that the Pentagon is completely incompetent, but maybe, just maybe, you should do some fucking research before your criticize somebody.

    Completely deleting data is very, very hard. Wiping a drive securely [auckland.ac.nz] against a determined and experienced foe may take more than 20 passes. Considering the physical security at most schools, giving the drives to schools is as good as handing them to the Russians/Chinese/Martians or whoever is the enemy of the day. If the drives haven't been properly wiped, you might as well give them accounts on all the DoD computer networks so they don't have to steal computers from school kids. I also would say it's a lot easier to smash a drive into itty bitty pieces and burn them than it is to properly erase it.

    When dealing with national security, one should generally err on the side of caution.

    Finally, for all you sysadmins and security experts out there, how would you like it if politicians with no computer knowledge whatsoever were second-guessing all your security decisions, while making sure you'd still take the fall if anything went wrong?
  • ... even the computers that had classified information on their harddrives should be allowed to be re-used.

    No, no, no! Information can be recovered long after the second or third overwrite. Here's my
    source [auckland.ac.nz]. And that's just what the public knows how to do. Who knows what the NSA and their foreign counterparts can manage.

    Remember that, when dealing with data security, you don't just need to worry about what your enemy can do now, you need to worry about their capabilities 20 years from now. If the data has to stay secure for 50 years, then the only choice is to destroy the drive (we may have nanotech by then, and then you'd probably be able to uncover everything that was ever written to the drive).

    Please, people, the NSA and the Pentagon have people who know way the fuck more about security than you or me. Leave matters of national security up to them, and go back to worrying about how to make your home linux box secure... Write to your politicians, and tell them to stop meddling. Getting 486s for schools may sound like a noble cause, but if it costs the lives of American soldiers 5 years from now, was it worth it?
  • It takes a lot of effort to securely erase the data on a hard drive. The ideal is to use encrypted FS from the beginning and then do multiple overwrites with random data to create a lot of noise, and even then recovery may be possible.
    My question is, how old are these systems being retired? If they are relatively old, then the hard drive is probably very low capacity by todays standards and cannot be considered too reliable.
    With a low-capacity hard drive, would it be cheaper to just ditch the hard drive entirely, using a destructive secure erasure method and spend the moeny that would have been to pay for staff to erase the disk without destroying it to instead buy a new small hard drive?
  • A small drive could hold LILO and /boot. Once a Linux or BSD kernel is running, the other drive should become accessible. Linux and the BSDs don't rely on the BIOS to access drives. So boot from either a floppy or a small hdd then load the rest of the system from the "big" drive.
  • And they still haven't fixed it! What I hate most is the way filename completion won't work with of=/really/long/path/here/and/really-freaking-long -filename.ext I gotta type the whole shit out.

    If you know how path expansion works in UN*X, a '*' expansion would be looking in your current (home?) directory for a subdirectory named 'of='. The unix shell does not know (or care) about what the command being called expects. That's why most UN*X commands do something like '-i filename'. That way the filename stands alone, and shell 'glob' expansion will work properly.

    If you want to avoid the whole problem, then you can replace

    dd various_options if=/some/long/pathname of=/some/other/long/pathname

    with

    dd various_options < /some/long/pathname > /some/other/long/pathname

    That way, the filename is standalone and shell glob completion will work properly

    In my early Unix days, I wondered why dd even had the if= and of= parameters, since < and > do the same things. I think it's so that people who get used to all of the other var=value params for dd stopped bugging the writers for a way of naming the input and output files that was consistent with the rest of dd's options. Perhaps we should update the man pages to reflect that.
    --

  • Not a big problem... If it's a one time pad, you rewrite the pad everytime you rewrite the data. That way, both sides always get 'random' writes.. Even if the data doesn't change, you can't tell, because both sides get re-randomized. You need both sides to know what's going on.

    The nice thing is that, since both drives are always getting 'random' data, a couple of extra passes of 'random' data might make it real hard to figure out which pass was the 'real' random data. Randomly switching which side gets the 'pad' and which the data-modified 'pad' may make analysis even harder.

    A prototype of this 'secure' filesystem could probably be done up in a couple of hours by modifying a software-raid driver. If i cared enough about my own security, I might do it myself.

    I'd patent this idea, but it's now been published
    --

  • My guess is that the 'classified' method of blanking a drive includes writing data multiple times, with a pause (weeks, months?) between. to allow magnetic data to 'bleed' into the borders. -- then, as someone mentioned, run it through a degausser to suck the platter dry and do a low-level format.
    (I'm guessing, here)
    --
  • the IF= and of= constructs do not open pipes. The dd program has to parse the strings and wilfully open the input and output files named.

    If you don't specify an if= or of=, then dd is capable of being used as part of a pipe.
    --

  • NO! A pseudo-random generator can be backtraced.
    /dev/urandom is not a pseudo-random generator, because it uses truly random data which accumulates across the kernel over time. Well, it does use a pseudo-random generator when there aren't enough truly random bits lying around, so if you're /really/ paranoid then you should use /dev/random, which doesn't. Nevertheless, it is much, much better than using an entirely deterministic rand() function.
  • Can't you just fire a couple of EMPs at the disk first? Or would that damage the hardware?
    ------
  • What about taking the platters out (in a clean room), zapping those several times, and replacing them?
    ------
  • "They" do. Hard drive capacities double every x months, and every data that was erased from a hard disk that was new at one point can be recovered y months later. There is a linear relationship between x and y. That's Moore's law.
  • IIRC, it's:

    1. Write all 1's (0xff) to the disk. Readback & Verify.
    2. Write all 0's (0x00) to the disk. Readback & verify.
    3. Repeast steps 1 and 2 two more times (for three full passes).
    4. Write a random byte to the disk (I'm partial to 0x47 or 0xb2, but take your pick). Readback & verify.
    5. Write the 1's complement of the previous byte to the disk. Readback and Verify.

    Make sure to generate a log somewhere for audit trail purposes!

    Note. I am NOT a DoD STD. Please check your customer's relevant specs for purge.

  • Re:dd is not good enough to erase data (Score:5)

    by sconeu ( 64226 ) on Sunday June 10, 2001 @03:16PM (#162343) Homepage Journal
    Having written several disk purge (not declassification -- once it's classified, it's classified) routines, I feel qualified to comment.

    You don't just "dd if=/dev/zero of=/dev/rhd0".

    There are several standards for purging media, such as DoD 5220-28M, and AFR 205-16. They specify the means of purging various media, ranging from core to disk to tape.

    The smartass comment about "dd" was totally unwarranted. Also, if hph had bothered to READ the article, he'd have seen that DoD was simply reverting to the earlier policy of wiping (unclassified) disks and donating the computers, instead of destroying all disks.

    Even the posters don't read the fscking articles any more.

  • It's a versatile UNIX tool that, among other things, can be used to write "zeros and ones" to an entire harddrive (...not just the filesystem)
  • I suspect that said windows licenses are non-transferrable. That is usually the case with corporate/govt. site license arrangements. As a result, it would probably be illegal to sell these computers without removing windows from the drives anyway.

    ________________________

  • dd is not good enough to erase data (Score:5)

    by Baki ( 72515 ) on Sunday June 10, 2001 @11:13AM (#162351)
    At least a simple dd if=/dev/zero of=/dev/hda is not. There are still ways (disassembling the drive and analyzing the magnetic surface carefully) to retrieve old data written before a constant bit pattern was written, and I'm sure the Pentagon does take such possibilities (difficult, but it can be done) seriously. A real good erase must write several different bit patterns in a row to make sure the original bits have been changed several times. The story isn't as stupid as some might think.

  • Re:Its only UNCLASSIFIED data... (Score:3)

    by flatrock ( 79357 ) on Monday June 11, 2001 @05:35AM (#162356)
    Confidential information makes it's way onto unclassified computers in the military. The people dealing with the information are human, and some information almost inevitably ends up on the computers that they use on a daily basis. Those computers are not intended to have classified information on them, but as people write reports, and prepare presentations, small amounts of information leaks into what they produce. Each report or presentation doesn't hold enough information to be a problem, but taken as a whole, a lot of information can be gathered.

    I did computer support at an Air Force base a number of years ago. The only time I heard of information not being properly destroyed was when a hard drive failed that had personel info on it failed, and the computer tech threw it in the trash when he replaced it. The tech had been around a long time, so he managed to not lose his job over the issue, but he should have known better.

    The Air Force policy where I was at was that a computer's life span was 5 years, and they rarely bought state of the art computers in the first place. After a few years, the departments with the budgets to buy new computers would buy them, and the still usable used computers would be passed to a department why didn't have money to upgrade their 5+ year old equipment. Often those 5+ year old computers would be passed down to contractors and others without the budget or political clout to acquire new or even slightly used equipment. Old computers were also scavenged to keep other old computers running. By the time the Air Force was done with those computers, there was very little value left in them, even for educational use. An average computer tech contractor costs the government somewhere in the range of $40 an hour. If it really worth spending the time to make sure the computer's drive is wiped. In many cases the computers don't even work, so wiping the hard drive means putting it in another computer to do so. In the end the schools get tons of junk which they have to pay to dispose of, and the government gets to be politically correct.

    Just chuck the hard drives in the incinerator and throw the computers away. Don't waste the time and effort trying to figure out if there might possibly be sensitive (classified or otherwise) information on the hard drive, destroy it. At the point the government is willing to give them to schools, they aren't worth anything anyway. If the government wants to spend money on getting rid of old computers, spend it recycling the old parts, so we don't just put them in landfills. Giving them to schools may be politically correct, but mostly it just means that the school wastes it's resources trying to find out if the computer is usefull, then just chucks it in a different landfill.

  • GPart (Score:2)

    by VB ( 82433 )
    You'd be amazed at what this tool [uni-hannover.de] can recover.

    Linux rocks!!! www.dedserius.com [dedserius.com]

  • The Gutmann Paper (Score:5)

    by Speare ( 84249 ) on Sunday June 10, 2001 @12:06PM (#162360) Homepage Journal

    For a fairly exhaustive paper regarding the secure deletion of data, see the Gutmann paper on USENIX.

    Secure Deletion of Data from Magnetic and Solid-State Memory [usenix.org] by Peter Gutmann

    This covers a series of 22 overwrite patterns that are formulated to ensure proper destruction of any trace information on RLL- and MFM-encoded hard drives. It goes into some detail about the ways electron microscopy may be used to recollect trace information. Other patterns exist, and I'm expecting the DoD or NSA has even more rigorous schemes.

    Unfortunately, raw degaussing of a whole hard drive device often disables the device's ability to operate in the future, or is not strong enough to ensure the destruction of the data.

  • ... even the computers that had classified information on their harddrives should be allowed to be re-used. Considering the fact that you can still get at second and third generation overwritten data, just overwrite 10 or 20 times with random junk (not a predictable pattern of 1's and 0's) and you've most certainly buried the classified stuff.

    If you REALLY are paranoid, just get a script to plant misinformation throughout the system before deleting it all.

  • Re:dd is not good enough to erase data (Score:5)

    by OmegaDan ( 101255 ) on Sunday June 10, 2001 @12:45PM (#162367) Homepage
    My question is -- if all this data can be recovered in 3rd or 4th generation wipes ... why can't they make a hard drive that exploits this to double its capacity ?
  • The intent of destroying the hard drives was to prevent sensitive (though technically unclassified) information ending up in the hands of someone who might have some fun with it (including an inquisitive High School student who finds data and decides to sell it to his druggie friends who.....).

    Please consider, as well, that calling it a 'preventive measure' is probably a euphimism for somebody found some useful information on a hard drive we gave away and three informants died as a result -- but we're not going to acknowledge that that's the real problem because it might end up in the whole program being trashed..

    You have to look at this from the (probably non-technical) Bureaucrat's point of view. Once the drive is destroyed, the problem is solved. Paying good money to buy hundreds (thousands?) of brand new hard disks that you're going to give away is a waste of funds that you can always find someone in your organization clambering for to help hunt down the latest killer. (remember that this initiative probably made it past on the bigwigs based on the promise that it would cost the department next to nothing, while providing good PR). Providing new drives with every outgoing machine would probably increase the up-front cost of the program by an order of magnitude.

    (the scarey thing is that the 'destruction' probably consisted of tossing the drive in the garbage where any spook would be happy to dumpster dive and retrieve all of this data from one place.
    --

  • Well if it's so cheap, the school should be happy to pay $20-$50 for a used drive to make a free computer usable. It's a far more justifiable expense for the school recieving the drive than it is for a government department that is going to be giving (thousands of) machines away as a (free) goodwill gesture.
    --
  • Back in the early '80s was someone posted a note explaining that when the NSA/CIA tested disk drives, they had a prescribed method for the vendor to overwrite the drive before it could be passed on for other uses.

    Most vendors found it easier to just melt down the drives.

    Remember that these were mainfraime hard drives, in the early '80s, that probably cost in the range of thousands of dollars each (retail).
    --

  • Unclassified doesn't mean unimportant. Your computer is unclassified, but I'm sure that I could get a whole lot of 'interesting' data about you, and your job, by looking at the data on your hard drive.

    Remember that these are 'unclassified' machines and they feel that the risk of 'only' overwriting them is fine. They're still destroying drives from classified machines.
    --

  • what about fill the disks with the bytes "6c61 6f6c 676e 7420 206f 7375 2820 6570 746e 206c 6874 7369 6220 7469 2073 6562 6761 6e6f 0a29"??

    Thats ascii for "all this bits belong to us (pentagon)"... : )

  • Actually, an acid bath is the generally accepted way to prevent a disk from ever being read again.
  • if they're worried about whats on unclassified machines, then somebody's not doing their job
    Remind me never to hire you for a computer security position. BECAUSE they're worried about what might be on an unclassified machine, they ARE doing their jobs. You don't stay secure by making assumptions. Period. You do it right, every time.
  • Is it me or hasn't anyone heard of srm [sourgeforge.net]?
  • A reply to a story you moderated renders all of your moderations undone....

  • OpenBSD disk are not physically secure. 1) Openan OpenBSD boxen (or 99.9% of the computers out there) 2) take drive out 3) mount disk 4) look at all your data and p0rn

    OpenBSD only encrypts swap (and you have to turn this on manually) and doesn't have support for encrpting ufs yet...

  • Unfortunately, unless you manage to completely pulverize the drive, parts of the platters will still be intact, and data can be read from them. So a hammer is not the right solution. But if you really want a mechanical solution, I would probably recommend a mill (grinding it to flour :-)

  • Re:LAYER the security approach (Score:3)

    by joto ( 134244 ) on Sunday June 10, 2001 @02:16PM (#162393)
    What if you layered the security approach? Encrypt the filesystem with a very good cypher and encrypt the entire filesystem!

    Please tell me how to make sure they used encrypted filesystems 5-10 years ago, on those old machines that they are scrapping now. (See..., it's not a very helpful suggestion!)

    Actually, it wouldn't help security very much to use encrypted file-systems either. Encrypted file-systems are only supposed to help if someone seizes your machine. That means it might be a good idea on a laptop, but if physical security is good, it is an unnessecary hassle to use on desktops. And, as avoiding loss of data is equally important when it comes to security, I wouldn't really think they would want to go through with that. Better just scrap the HD's.

  • Sorry this is not correct. The reason I know this is that I used to work in a Secure Computer Information Facility or SCIF in military speak. Part of this meaning in English is that you work in a *lead lined building* and that every *square foot* of soil (for some odd reason) has been certified for said installation. It also means you cannot bring radios, programmable watches, calculators, (anything with memory) to work. If you think this is fanatical, any media which is taken out must be *stringently* degaussed, meaning you place a magnetic tape through a degausser at least three times, and as the article states (correctly) you *must* remove and degauss all disk platters muliple times. In fact, in my unit (a comm unit) before people pcs'ed (permanent change of station) they would give these platters mounted on wood with the unit insignia and your name on a brass plate. They actually look cool. (These were ancient VAX platters This was a rare exception as most items are destroyed, probably because the ancient aluminum platters kept cheesing the shredder.) What you are referring to is the writing of info at least *5* times over the *whole platter* with I believe F8/F0 hex which is a standard (albeit used for *unclass* and unclass EFTO (Encrypt for transmission only) drives. And there is NO way you could have recovered that with PC Tools. To sum it up: All classified media is destroyed, unclass could be saved, but a pain and generally destroyed as well.
  • hard drives can be read after simple deletion.
  • Ok, so even after 10 over-writes, there's still a chance that information can be recovered.

    What if you layered the security approach? Encrypt the filesystem with a very good cypher and encrypt the entire filesystem!

    Then, when you format the hard drive, overwrite 10x with dd (and random bit patterns, in a randomized write pattern), the black hats would have to
    1) Read through 10x overwrites,
    2) Read through the format,
    3) Decrypt the final result, protected with a strong cypher.

    As with all things security related, you provide multiple layers of defense!

  • Re:Whats dd (Score:3)

    by PurpleFloyd ( 149812 ) <`zeno20' `at' `attbi.com'> on Sunday June 10, 2001 @10:58AM (#162406) Homepage
    'dd' is a command in UNIX/Linux systems that allows direct writing to the disk bit by bit. Using a program such as dd many times over would make it nearly impossible to recover old data from the drive (when deleted with 'rm', files can be recovered by programs readily available on the market-- they are still on the drive, but the OS doesn't know where they are). When every bit is set to zero several times over, however, then there is no way in hell you will find old data on the drive.
  • It seems obvious once again that people are making snap judgements with insufficient information. The Pentagon's Public Affairs Offices are not exactly the prime repositories of technical ability. God forbid.

    dd would only perform a single write, which is in no way enough to delete all traces. That would be like erasing an audio tape; tiny vestiges of the original recording would still be present.

    The military has always known how to properly purge disks, for example with software such as Fortress, which has been around forever. It performs multiple passes of the disk, writing on the disk with different data, and in different scan patterns, with various read tests in between. It is very thorough. And let us not forget to mention large-scale disk degaussing.

    This "news" is much like the press "discovering" TEMPEST (the ability to sense radiations from a computer and monitor from a distance)last year, even though that technology has been publicly known for many years now.
  • Even with old Amiga floppy disks there were ways to retrieve some bits of data after a full format or overwrite. Blocks are separated by small gaps and tracks don't take up the whole 360 degrees of magnetic surface. On the Amiga, the disk controller was optimized for speed so it would not bother waiting for the index signal before starting to rewrite a track. Each time you rewrote a track, it would occupy a different section of the circle, leaving some old data where a clever guy could still retrieve it.
    Today's hard drives are much more sophisticated than this, so they sure leave many more chances open to retrieve old data in original ways.
    //BernardoInnocenti

  • Heh (Score:2)

    by IainMH ( 176964 )
    "The overwriting entails printing series of ones and zeros over the stored material".

    I bet that someone at the Pentagon is now looking at the first two buttons on thier keypad and thinking;
    G-ddamn this is going to be boring
    :-)

  • Uh, it's the reporter, not the Pentagon that claimed that the Pentagon has "found a way" to erase the hard drives.

    Is it just me, or does it seem to anyone else like Slashdot's editors 1) can't read, and/or 2) are easily amused?


    --

  • Re:Politics and Security don't mix (Score:3)

    by Erasmus Darwin ( 183180 ) on Monday June 11, 2001 @05:22AM (#162422)
    True, although I have to say that "national security" is the biggest lie that ever was told. It's been used repeatedly to engage in things so immoral that I just want to puke.

    Amen! I hate it when the DoD uses the argument of national security to rationalize nothing less than the wholesale MURDER of thousands of innocent harddrives. The poor little devices served that country well, by storing classified data, and their loyal service was repaid with incineration.

    Even the brave harddrives that assisted with unclassified work still had their brains wiped clean several times before being forced into the hellish public school system. It's barbaric.

  • Thanks - I see someone else modded it up as well. For some reason I seem to get this more than the average user, from what I can tell (modded 'troll' for no apparent reason). Good to see the system eventually rights itself. :)

  • Windows licenses (Score:5)

    by mgkimsal2 ( 200677 ) on Sunday June 10, 2001 @01:03PM (#162431) Homepage
    Everyone's talking about 'dd' - to my knowledge it's not a Windows utility. My hunch is that most of these computers are Windows systems, meaning we (taxpayers) have paid for a license. Unless the gov't has immaculate record keeping and can provide the license for Windows (and the media to restore it) the recipients of these systems are going to pay again for a Windows license. True, not everyone will put Windows on, but my guess is that a majority of them will be put back in service with Windows on them.
  • 'dd' is a command in UNIX/Linux systems that allows direct writing to the disk bit by bit.
    Not quite. It allows direct writing to the disk controller. The magnetic pattern written to the physical disk is (unimaginatively enough) controlled by the controller, which tells all sorts of lies to the computer. For example, the controller may detect that a sector can't be read back to contain the same data that were just written, and transparently move the data to a substitute sector without ever informing anyone or anything of this decision.

    The program that can see past the controller's shenanigans is called "low-level format". It is more akin to a device driver, having intimate knowledge of how the actual disk operates. If the Pentagon wanted to contract with the manufacturers of disk drives for a very special LLF utility that properly exercised each sector, writing magnetic patterns specifically designed to cripple subsequent analysis, that might be good enough.

    Or it might not. You've got to ask yourself how much effort an attacker is willing to expend to retrieve the contents of that drive, and how much damage can be done if he's successful to properly evaluate the risk.

  • New Classroom Projects (Score:4)

    by corvi42 ( 235814 ) on Sunday June 10, 2001 @01:11PM (#162447) Homepage Journal
    Just imagine the new educational opportunities this opens up: young school children can now write letters to undercover agents, special forces members, and secret government think tanks.

    They can also gain a valuable jump-start on children in other countries by starting young to learn about data analysis and retrieval, surveillance and the ins-and-outs of the military-industrial complexe.

    This could be the best strategy to educate american children in the face of the growing espionage-publicity gap caused by the recent spade of british agents "losing" their laptops.

  • Until it's your name/job/reputation on the line. In the absence of a business model, most government folks of any strip play a conservative hand, unless there is good press involved for doing something intelligent. It's a fear driven, inefficient culture. David Gergen said this about it, recently:
    But that's just the tip of the iceberg. The civil service-the substructure that is so vital to day-to-day operations-is rapidly crumbling: 53 percent of the federal workforce will qualify for retirement by 2004; 71 percent of the government's senior managers can retire by then.

    And there's precious little new blood to replace them. A survey of the nation's most academically gifted college students-the Phi Beta Kappa graduates-found that only 1 in 10 rated the government as the employer of choice. Among the nation's public-policy schools, interest in government has also declined. According to Light, some 76 percent of those graduates sought public-sector jobs in 1973; two decades later, the number dropped to 49 percent; today it has dwindled to about 30 percent. And these are people supposedly being trained for public service!

    in US NEWS [usnews.com]

    There is a temptation to blame recent moral turpitude in elected officials for the intellectual vacuum of the civil service corps, but calling it a leadership issue is oversimplifying.

    Closer to the mark, we get what we pay for. I just turned down a GS position, because the pay was 2/3 a private sector offer.

    Go figure why we are shocked by this decision to spare unclassified hard drives.

  • Department of Defense
  • This is wrong. You cannot obtain personal files of random Pentagon employees with the FOIA. Yet, these could be exposed by giving away an old hard drive.

  • not very interesting (Score:4)

    by flynt ( 248848 ) on Sunday June 10, 2001 @10:56AM (#162458)
    All this is is a short 2 paragraph write up on policy of giving away old hard drives to schools that might contain senstivie (not classified) data. No where does the pentagon claim to have discovered anything new here, and the way it is written in the story is just to explain it to people who don't understand that information might be able to be retreived from a hard drive. If anything, this is only interesting because it marks a slight change of policy, but beyond that it seems the only reason Taco posted this was so that he could show that he too can find some (however minor) faults in a CNN story's technical content. So to answer your question, no, this story did not amuse me as much as it apparently did you.

  • Now, I believe (and if I'm wrong, correct me), that when a bit is written to the hard drive, there is a minor amount of error where it could be written to physical area of the hard drive. The error isn't enough to make it interfer with the other bits of data on the hard drive, but is enough that there is a decent amount of chance that you could pull off the previous value of the bit from the hard drive, if you tried hard enough.

    Also, when a bit is changed from one state to another, if the bit is changed, it might vary slightly (but still in the range of tolerance) if the bit was changed from 0 to 1 or 1 to 0 instead of the bit being 1 before or 0 before and remaining unchanged. Think of it as the magnetic field defining the bit as being not as strong as if the bit was not "flipped".

    Now, the way to get around this is to write random bits of data to the hard drive several times. Another poster has commented that seven is the magic number, and since I've heard that number before, I have to agree. So fill the hard drive with random bits seven times and the original data can be assumed destroyed.

  • In other news... (Score:4)

    by Edgewize ( 262271 ) on Sunday June 10, 2001 @10:57AM (#162463)
    Pentagon officials today reversed a six-month-old policy that stated that used whiteboards must be pulvarized with sledgehammers before being thrown out or given away. This move allows whiteboards to be donated to classrooms.

    Deputy Secretary of Defense Paul Wolfowitz is credited with discovering that nonclassified material could be removed from the whiteboard with an eraser.

    An anonymous source close to the Pentagon has stated that this is undisputably the smartest decision the government has made in years.

  • Re:Politics and Security don't mix (Score:3)

    by dario_moreno ( 263767 ) on Sunday June 10, 2001 @12:03PM (#162465) Journal
    true ! A few years ago, in France (ok, not the best example for clever militaries, but where are they to be found ?), a satirical journal, le Canard, managed to get an almost complete list of the military secret service : they got hand of some class reunions invitation lists of military schools, and the secret service men were listed as belonging to units that did not actually exist.
  • Arguably, Norton has this feature with any program ::grin::.

    Just install it on any Windows machine and it'll do its job within a few days. :P

  • That could work, but I think you could buy a new (probably 10X bigger) drive for less. You can also bulk-erase _through_ the aluminum drive case, but modern hard drives require a very high magnetic field strength to erase (several times that required for the strongest ("type III") magnetic tape). It would be likely to bend or break the platters. And finally, there is no mil-spec qualified bulk-eraser for even type III tape--so regardless of technical merits, they couldn't use bulk-erase methods.
  • One more thing: Since the heads don't follow the exact same path every time, you've also got to do something to ensure that the fringe areas out to each side of the track really got overwritten. If you can get direct control of the head position microstepper, you write once a little to the inside of the nominal track position, then write once a little to the outside. Repeat for the other 22 patterns required to be sure of complete erasure... It's a big pain to write the program, but once written you just put it on bootable DOS floppies with the instructions: "Insert in disk drive, turn power on, wait until complete (up to 3 days)." A good link to a thorough discussion of this has been posted before.

    Of course, there is a big assumption when erasing this thorough is required: that someone will get hold of the drive and believe that it is worth a great deal of work by high-paid techs using very expensive test equipment in a clean room. If the computer has a sticker saying "War plans room. May contain classified data", there's a pretty good chance someone would do that. But if you simply dd (or a DOS/Windows equivalent) a shipment of 100 machines scrapped from the personnel dept, do you really think anyone is going to dissassemble all 100 hard drives on the chance that maybe one of them contains useful data? (And don't you think that the school receiving those machines might investigate who's been stealing the hard drives?)
  • Is any of that information unavailable through other channels (budgets, FOIA requests)?
  • I have an example of a legitimate 20 year secret. The F-117 was prototyped in 1977 as the "Have Blue". We non-gov't pinkboys still do not know the radar cross-section and full capabilities of this aircraft, to the benifit of national security. The same thing can be said of aircraft carriers, but to a lesser degree.


    Brant
  • [recounting of vague memories of secure deletion requirements]

    [absurdly thorough overkill method for secure deletion]

    [suggestion that previous method is barely adequate]

    [expression of smug superiority]
    --
  • [accusation of incompetence reiterated]

    [inadequate deletion method presented as obvious solution]

    [insultingly simple and inadequate recommendation for general solution to computer incompetence]
    --
  • My high school gets all the 486s and low-end pentiums it can use from local businesses.

  • unclassified machines

    If somebody hasn't already pointed it out, unclassified data can be combined to derive classified data. Say a terrorist already has the following information.
    • Uranium-235 is used in nuclear weapons. (publically available)
    • Tritanium shielding can be used to mask radiation from large amounts of Uranium-235. (publically available)
    • A military truck from Utah carries supplies along Interstate 12 every few months. (spy)
    Nothing too dangerous in that, is there? But what if he discovered this:
    • A vehicle production facility in Utah ordered twelve tons of tritanium. (unclassified)
    From that, a terrorist could theorize that he could aquire purified Uranium-235 for production of nuclear weapons by hijacking the military transport trucks running along Interstate 12. Clear?

  • Re: several times over? (Score:5)

    by F00Fmaster ( 455601 ) on Sunday June 10, 2001 @11:57AM (#162509)
    Seems a little funky

    Yes, it is very, ah, 'funky'. You can use magnetic analysis of the drive to get back data written to it a long time ago. Basically, the only way to actually remove data from a drive is this:
    • Overwrite about a dozen times with pseudo-random noise, not just zeros. Simply overwriting with a constant pattern (just zeros [000...] or just ones [111...], or just a pattern [01010101...]) is easy to read through in magnetic analysis. You can do this from the command line: dd if=/dev/urandom of=/dev/hda
    • Use a cryptographically secure number to write truly random data over the drive, to which no pattern can be found. This is the hardest step, and it should be done several times.
    • Finally, wave a magnet over it to scatter the remaining magnetic field. This shouldn't really be so much a 'wave' as a 'continuous bombardment that goes on for several hours', but the idea is the same.
    • Then, finally, you also format the drive. Simply formatting it or simply running 'dd' does nothing to prevent the people the Pentagon is afraid of from getting the data.
    It isn't so much about reading back single bits, but about reading whole files, in which a single bit or two might be slightly damaged. For example, read the following sentence:
    Th` new b`mber is m`de of tritanium oxid`.
    The meaning of the sentence remains intact, even when four or five characters are lost. In the same way, quite a great amount of data can be uncovered by reading large chunks of data, even if you can't retrieve everything to the bit.
    Making these decisions based on politics and not security is a dangerous choice. I hope the Pentagon thought about it very seriously. Of course, with the Presidents virtually giving away our nuclear missile designs, there isn't much left to protect, but still the issue stands.

Slashdot Top Deals

He has not acquired a fortune; the fortune has acquired him. -- Bion

Close