👽

  • TheBananaKing@lemmy.world
    link
    fedilink
    arrow-up
    30
    arrow-down
    1
    ·
    5 months ago

    Most of these answers are mostly right: deleting a file on disk doesn’t actually erase the data, it just marks the space as available to write over - meaning that so long as nobody’s used the space since, you can go retrieve the contents with an undelete utility.

    Most of the time, people don’t care - but if for instance you’re selling the PC or there’s highly sensitive information involved, that might not be good enough.

    As such, there are utilities that can go out and specifically overwrite the contents of a file with all zeroes, so ensure that it’s dead-dead - and there are other utilities that can do the same to an entire disk.

    There’s one wrinkle: Magnetic HDDs don’t reliably erase and overwrite completely in a single pass; just like rubbing out pencil writing, it can leave faint impressions under the new content, and it is actually possible (with serious effort by forensic recovery people) to glean some of the previous content. If there’s serious money / security at stake, a simple overwrite is not enough, so there’s software that certifiably-randomly scribbles over each bit, seven times over, making the chances of recovering the original astronomically slim. Again, this can be done for individual files or the entire disk.

    SSDs aren’t prone to leftover impressions, thankfully - what’s gone is gone. And they have one other neat feature: while a magnetic disk can only be erased one bit at a time, so large disks can take hours - SSDs can just open the floodgates and ground every cell at once, fully erasing the entire disk in an instant.

    This instant-erase, while comprehensive… returns before you’ve even taken your finger off the ENTER key, so fast it feels like it can’t possibly have done anything, it must be broken, how can I trust it? So BIOS manufacturers hype it up, call it something impressive to underline that it’s big and powerful, and actually impose a 10-second countdown to make it feel like it’s doing something complicated.

    Any of these different things have been called ‘secure erase’ at various points, so it’s a little context dependent. But from the end-user perspective: this data is getting shredded then incinerated then added to cattle feed; it’s not coming back.

    • listless@lemmy.cringecollective.io
      link
      fedilink
      English
      arrow-up
      17
      ·
      5 months ago

      THIS IS THE HILL I DIE ON.

      No one has ever recovered overwritten data, as far as anyone can tell. Go look it up. The technique was only a theoretical attack on ancient MFM/RLL hard drive encoding (Gutmann’s paper). Even 20 year old drives’ (post 2001, approx) magnetic encoding are so small there isn’t an ‘edge’ to read on the bits. A single pass of random data is sufficient to permanently destroy data, even against nation-state level actors. Certainly enough for personal data.

      from https://en.wikipedia.org/wiki/Gutmann_method :

      Most of the patterns in the Gutmann method were designed for older MFM/RLL encoded disks. Gutmann himself has noted that more modern drives no longer use these older encoding techniques, making parts of the method irrelevant. He said “In the time since this paper was published, some people have treated the 35-pass overwrite technique described in it more as a kind of voodoo incantation to banish evil spirits than the result of a technical analysis of drive encoding techniques”

      More reading material:

      NOW THAT BEING SAID there is no harm in doing a secure, 35-pass overwrite other than the time, energy and disk wear. If watching all the bit-patterns of a DoD-level wipe using DBAN on a magnetic disk tickles your fancy, or you think this is a CIA misinformation campaign to get people to do something insecure so they can steal your secrets, please just go ahead and do a 35-pass overwrite with alternating bit patterns followed by random data. I can tell you that I believe in my heart-of-hearts, that one pass is sufficient.

      • FuglyDuck@lemmy.world
        link
        fedilink
        English
        arrow-up
        12
        ·
        5 months ago

        This is exactly what a cia analyst whose tan literally comes from their monitors and is never let out of Langley’s 38th sub basement would say.

        Guess we’re doing 40 passes. Just to be sure. ;)

      • jet@hackertalks.com
        link
        fedilink
        English
        arrow-up
        3
        ·
        5 months ago

        In my industry we destroy all storage devices when computers are returned at end of lease, or decommissioned.

      • TheBananaKing@lemmy.world
        link
        fedilink
        arrow-up
        3
        ·
        5 months ago

        Interesting. We mostly use DBAN at work because it’s a one-button process you can walk away from, and it has drivers for hardware old enough that we’re disposing it. Nobody’s ever selected the fancy super-paranoid stuff as far as I know.

        If the hardware won’t boot, we take a layer-1 approach instead :D

      • RizzRustbolt@lemmy.world
        link
        fedilink
        arrow-up
        3
        ·
        5 months ago

        When I worked at the e-waste recycle and technomancery place we’d do secure wipes for any hard drives they dropped off with their stuff.

        And one time somepne asked if we could do a Gutmann wipe for his hard drive.

        His 10TB hard drive.

    • KISSmyOSFeddit@lemmy.world
      link
      fedilink
      arrow-up
      2
      ·
      5 months ago

      There’s one wrinkle: Magnetic HDDs don’t reliably erase and overwrite completely in a single pass; just like rubbing out pencil writing, it can leave faint impressions under the new content, and it is actually possible (with serious effort by forensic recovery people) to glean some of the previous content.

      No, it’s not. At least it has never been done on more than single bits, with an accuracy higher than 50% (which means even when trying to restore a single byte, your chances of it being correct are 1/256).