[clug] Secure Erase of disks/drives

Alex Satrapa grail at goldweb.com.au
Wed May 9 00:28:40 GMT 2007


On 08/05/2007, at 20:09 , Michael Cohen wrote:

> It just makes sense - if there was an easier way to read say 1 or 2  
> time
> overwritten data (say by some great dsp technique/better heads etc)  
> - your
> driver storage could just be increased by a factor of 2 with little  
> cost.

The whole idea of scanning the disk with a ultra-sensitive read head  
is that you can make specialist equipment that is 10 times more  
sensitive than the mass-produced stuff. In addition, the drive is  
built to recognise eg "0.78" as "1", and "0.34" as "0". The more  
sensitive equipment is exponentially more expensive to build,  
maintain and operate. That is why it's not used in mass-produced  
consumer hardware.

So yes, if there was "an easier way", we'd have higher density  
drives. Two years ago I bought a consumer-level device which gave me  
500GB in a raid-0 enclosure. This year that model has been replaced  
with a 1TB version.

When consumer hard drives are capable of recording details on a disk  
with magnetic domains the size of one molecule, then I'll believe  
that it's impossible to retrieve overwritten data. Until then, the  
question will still be, "does any of the data that was ever stored on  
this disk pose a risk of harm if someone was to find it?" -  
otherwise, you are not paranoid enough.

When you are not paranoid enough, you make silly assumptions such as,  
"noone would ever steal that hard drive full of personal details of  
war veterans, because it's sitting on a desk in an important person's  
office, in a secure building - therefore we don't need to worry about  
physically securing the drive or encrypting the data."

Alex



More information about the linux mailing list