Thanks!

Posted by Kromey at 12:19pm Apr 10 '13
You must sign in to send Kromey a message

But wait, if one overwrite does the job, why is it argued that dozens are needed?

Short version? Because Gutmann said it was before he said it wasn't.

Actually, it stems from a fundamental misunderstanding of his paper, which does argue that data remanence is an issue for magnetic hard drives. Except, what it actually says is that hard drives using RLL(1,7), RLL(2,7), or MFM physical encoding need up to 23 passes each (the "35-pass Gutmann Method" being itself a fundamental misunderstanding of his paper) to thoroughly wipe them clean against a laboratory attack using a magnetic force microscope to "peel away" the magnetic "layers" on the disk to read what was written previously.

His paper also says that the data density and the complexity of "new" (in the mid-1990s) PRML drives makes the whole MFM attack useless from the start; the same goes for EPRML drives.

Your drives today? They're all EPRML (the magnetic ones, that is). RLL and MFM drives are long extinct, as they couldn't even get anywhere near the data density we've got today.

So basically what Gutmann said is that certain drives of a now-extinct variety need up to 23 out of 35 possible overwrites in order to clean them. That has been misinterpreted to become all modern drives need the full 35 passes.

There's even reason to doubt Gutmann on what he actually said in his paper, however. Remember Nixon? Remember the infamous 18½-minute gap? That tape is still around today, and despite having a lower data density than even the RLL and MFM drives Gutmann was talking about (yet using a similar encoding scheme) the lost data on that tape has never been recovered, despite multiple laboratory attempts and the fact that it was only overwritten a single time (well, as many as 9 separate re-recordings to add up to 18½ minutes, with undoubtedly some overlap, but at least most of that gap is just a single overwrite).

In fact, there is not one case of data ever being successfully recovered from any magnetic media after it has been overwritten as little as one time. Gutmann was talking about a hypothetical attack that is good in theory, but which even 20 years later we haven't been able to pull off.

Also, how about taking a strong magnet to the drive?

In theory, absolutely possible, and effective.

In reality, you have to remember that there is already a mighty powerful magnet inside the drive to control the arm. A magnet powerful enough to wipe the drive from outside the case would also be powerful to rip the iron straight out of your body -- so I don't recommend that. ;-P

However, if you pull the drive apart a simple neodymium magnet in direct physical contact with each platter is enough to wipe them. It's part of the process DiskStroyer uses, in fact -- just before you take sandpaper to the platters to also physically destroy the data tracks! (DiskStroyer is remarkably thorough for costing less than $40 for the re-usable (and rather useful beyond disk destroying!) kit and 10-minute process.)