On Sat, May 17, 2014 at 12:45 AM, ToddAndMargo <[log in to unmask]> wrote:
> On 05/16/2014 09:40 PM, John Lauro wrote:
>>
>> Also, be aware that unlike hard drives, for flash drives it
>
>> likely does not "erase" the drive. Internally the drive might
>> just mark the sectors as all clear and flag the previous contents
>> as erasable without actually clearing the flash. With a highly
>> over provisioned enterprise drive it may wait to actually clear
>> the flash chips. Not that the original data would be easily
>> accessible (probably have to take the drive apart), but it is
>> something to keep in mind if there is sensitive data...
>
> So "zero" does not clear the charge?
> I always "smack" my old drives with a hammer before tossing them.
It's fast and pretty effective, as far as the kernel knows. The
difficulty comes when what is actually on a drive is hidden behind the
controller, which has not yet bothered to actually write the "zero"
signals to the physical medium. And flash drives do some interesting
things to mark sectors as "quesionablel" and stop using them if
they've been written to too many times, and to load balance. So it's
an interesting question of someone with the manufacturer's detailed
specs could possibly read data off those other sectors. Such tools,
like tools to trace magnetic impressions off a hard drive platter, are
doubtless quite proprietary and expensive.
I suspect for any reasonable level of data recovery, 'dd if=/dev/zero'
is a very efficient and effective drive clearing technology. If you
feel the need to get enthusiastic, there is the old "shred" program
that does zeros, ones, 101010, 0101010, simi random data, etc., etc.,
etc.
|