You know how that data breach happened? Three words: eBay, hard drives

Re: Has it been six months already?

Yup...

And I still advocate chopping them in half... Oxy Acetylene being my weapon of choice.

Re: Has it been six months already?

Uhmm... dban and the like don't actually work. It won't erase data in the file system journal, or in bad sectors on disk.

Re: Investing

Yeah, that free DBan download kills most budgets, as does just simply folding most 2.5" drives in half and then peeing on them and burying them out in the rose planter.

Re: Investing

Yeah, like the free secure erase algorithm that is built into every disk drive controller since the end of the previous century? Activating that routine will really break everybody's IT budget.

Re: 10% ?

My initial reaction was that if the figure was 10% then "FORMAT C: must have been counted as a secure data erasure method.

On the other hand, perhaps the only people who think it is worth trying to sell old (and therefore slow, small and knackered) hard drives on eBay are bean-counters and *they* actually do have access to people with the necessary skills. Perhaps 10% is really true.

Edit: And if bean-counters are the only people selling then (for the same reasons) perhaps data thieves (and researchers) are the only people buying them.

Re: 10% ?

Realistically, how many folks have the ability to retrieve any data from $DISK following

# dd if=/dev/zero of=/dev/$DISK bs=4M

Yes, that is not a secure erasure technique. Yes, anyone with a few grand to spare might be able to convince a data-recovery company to retrieve some random fragments of what was there before.

More secure, if you care: download DBAN. Not sure that is officially secure either, because it lacks any bureaucratic certification of secure-ness. But it is the officially secure algorithm.

Re: 10% ?

Considering how many people on IT related web sites always suggest using dban, shred, dd and other utilities that don't work properly since they don't erase data between tracks or in bad sectors, I am surprised that that 10% were done right.

Magentic solution will destroy the drive as it'll wipe the all important servo data (which may or may not be desired), physical which destroys the platters is preferred and, weirdly, thermite isn't actually that effective (there are videos showng the residue being wiped off the platters) unless you're lucky/careful to place it right.

By far the best way is shredding.

But the IT dept trainee can't supplement his income if you make the disks unsaleable

Physical destruction is best

This will put the data beyond reach of anyone except a three-letter agency (and probably also the agencies).

0. Make sure it's a magnetic disk not an SSD or hybrid

1. Smash the electronics board with a hammer (probably optional, but satisfying).

2. Drill several holes in the top of the HDA

3. Put the disk in a tray and pour xxx-cola into one of the holes until the HDA is full, then more to cover it. (The multiple holes are to let the air / gas vent).

4. Leave overnight to dissolve the magnetic domains off the platters. (You know what it does to a tooth, right? )

5. Throw it away.

For an SSD you need an incinerator. (Or a decent bonfire, and utter disregard for the anti-pollution regulations).

Hacking a disk that's been 100% written to zero.

To get data off a disk that has been written to zero you need to hack at the hardware level.

When you write to a disk the head is not always exactly centred on the track. Sometimes it is off a bit to the left, sometimes off a bit to the right. So there may be a smear of previous contents as a weak noisy signal, if you are able to command the head to offset to various normally non-commandable positions left or right of nominal centre, and pick up the analogue signal from the head for nonstandard processing rather than feeding it into the standard disk-read signal processor code.

A three-letter agency might even have something like a large electron microscope to image the magnetic state of every square nanometer on a platter, and something like an image-processing system to decode it.

If you write multiple garbage patterns the chances of any off-centre data remaining goes down. I imagine each pass trashes about half of what was left by the previous one.

There is also whatever is left on the bad blocks that were replaced during the disk's lifetime. You have to assess what is the chance of a random four kilobytes written at a random time in the past being of any interest, and what might the consequences be? I'd hope that a disk's firmware erases a bad sector before relocating it, but unless the manufacturer specifies that it does you should assume the worst.

Re: Formatting has two options

You load a special driver into the drive controller that shifts the servo off track by 25%, then read the data left over between the tracks.

If you really want to erase a disk do this:

# hdparm --security-set-pass user /dev/sdX

# hdparm --security-erase user /dev/sdX

That will overwrite the data over the whole disk surface, on the tracks and in between the tracks.