| 1 |
On 17 Dec 2009, at 13:40, Marcus Wanner wrote: |
| 2 |
> On 12/17/2009 6:42 AM, Mick wrote: |
| 3 |
>> On Thursday 17 December 2009 05:13:32 Joshua Murphy wrote: |
| 4 |
>> |
| 5 |
>>> chicane ~ # shred test/ |
| 6 |
>>> shred: test/: failed to open for writing: Is a directory |
| 7 |
>>> chicane ~ # shred -v -n 25 -z -u ~/test/ |
| 8 |
>>> shred: /root/test/: failed to open for writing: Is a directory |
| 9 |
>>> |
| 10 |
>> shred ... shreds files. Therefore you may need to point it to the |
| 11 |
>> files in |
| 12 |
>> question for it to work. I suspect that if you point it to a |
| 13 |
>> device alone it |
| 14 |
>> just shreds the file representing the device on the Linux fs in |
| 15 |
>> question. |
| 16 |
>> |
| 17 |
> That would be a bit inconvenient...I still vote for dd, overwriting |
| 18 |
> the thing 26 times sounds like WAY overkill for a hdd... |
| 19 |
|
| 20 |
The US military specification is to overwrite randomly 3 times: http://en.wikipedia.org/wiki/Data_erasure |
| 21 |
I think the `shred` on the current System Rescue CD defaults to this. |
| 22 |
|
| 23 |
The advice to overwrite 26-35 times is, I think, based on Peter |
| 24 |
Gutmann's 1996 advice, which is now quite dated and is widely |
| 25 |
considered no longer relevant. Fair play to Gutmann: there aren't many |
| 26 |
studies on secure data removal made publicly available, so it was the |
| 27 |
best knowledge we had at the time. It may be accurate to the kind of |
| 28 |
drives available then, but not to those available now. |
| 29 |
|
| 30 |
Why not use dd? Grant says that his data "isn't too sensitive", so it |
| 31 |
doesn't really matter. But it's no more difficult to run shred than it |
| 32 |
is to run `dd` - it's about the same amount of typing. You might as |
| 33 |
well "do things properly" (also known as "following best practices"), |
| 34 |
even if you don't think you need to. 3 writes really doesn't take that |
| 35 |
long. |
| 36 |
|
| 37 |
Stroller. |
Archives