Wiping hard disks and company data
Discussion
I was reading an article in PC Pro at the weekend saying that X% of companies don't wipe their hard disks before ditching an old PC, and X% only overwrite it once.
Now I've heard this before, but surely if you overwrite something on magnetic media, it's gone ... a bit is either 0 or 1 isn't it ?
It is possible to tell what state a bit was in, say, 2 or 3 writes ago, and if so how (from a technical/scientific point of view). I'm curious
Now I've heard this before, but surely if you overwrite something on magnetic media, it's gone ... a bit is either 0 or 1 isn't it ?
It is possible to tell what state a bit was in, say, 2 or 3 writes ago, and if so how (from a technical/scientific point of view). I'm curious
Depends how its wiped.
If its a quick format the indicator is just reset, all the data is still there but the access to it has been removed.
Basically completely wiping a HDD requires a complex procedure to ensure that all the internal files that structure the data have been completely destroyed.
If its a quick format the indicator is just reset, all the data is still there but the access to it has been removed.
Basically completely wiping a HDD requires a complex procedure to ensure that all the internal files that structure the data have been completely destroyed.
nevpugh308 said:
Now I've heard this before, but surely if you overwrite something on magnetic media, it's gone ... a bit is either 0 or 1 isn't it ?
Ah ha, but we don't live in a digital world (ok ok let's ignore the quantum stuff for now). The 1's and 0's are obviously stored as magnetic patches on the disc. Even if you overwrite a 0 with a 1 it doesn't totally obliterate the 0. Let's say the 1 is really 0.97.
The disc drive doesn't care about this, but My man with his magnetic probe does.....The upshot being, even if you format (full) drives you can get the data back with enought time / money.
Write the disc with multiple passes of random data and you're pretty safe however.
In short, yes.
Which is why the US department of defense format standard is to overwrite the data 7 times.
Essentially (IIRC) there is a 'ghost' charge present that can be read; the data may be being read as '1', but careful measurement shows it is actually 1.0000001, and the inference is that prior to being written with '1', it was already '1' or '0'. If that makes sense. It's a bit like dtmf tones; both can exist in the same place at the same time.
Which is why the US department of defense format standard is to overwrite the data 7 times.
Essentially (IIRC) there is a 'ghost' charge present that can be read; the data may be being read as '1', but careful measurement shows it is actually 1.0000001, and the inference is that prior to being written with '1', it was already '1' or '0'. If that makes sense. It's a bit like dtmf tones; both can exist in the same place at the same time.
Phil : So are you saying that, in simplistic terms, if a bit is 0 then you overwrite with a 1, the actual value will be 0.97. Overwrite again and it might go to (say) 0.99, then a final time it might go to 1.0 ?
Is that a constant, or might it sometimes go
0 > 0.97 > 0.99 > 1.0
and sometimes
0 > 0.95 > 0.97 > 0.98 > 1.0
or sometimes
0 > 0.99 > 1.0
and so on ?
Is that a constant, or might it sometimes go
0 > 0.97 > 0.99 > 1.0
and sometimes
0 > 0.95 > 0.97 > 0.98 > 1.0
or sometimes
0 > 0.99 > 1.0
and so on ?
nevpugh308 said:
Phil : So are you saying that, in simplistic terms, if a bit is 0 then you overwrite with a 1, the actual value will be 0.97. Overwrite again and it might go to (say) 0.99, then a final time it might go to 1.0 ?
Yes...and (you knew this was coming) no. Think more of the analogue side of whats happening. The disc gets to a certain point and the head coil fires. It arranges a good number of the magnetic poles the correct way for a 1. However it :-
1) May not set them all.
2) May not fire at the same place.
3) May not fire for so long (or longer).
If you were ever able to see the layout of the disc you'd be surprised how inaccurate it is.
So, in general, writing continuous 1's may push you that way...but not always all the way (usually due to 2).
The reason you write random 1's and 0's is even though you can detect a pad is 0.997 you have no idea now that it may have been a 0 > 0.991 > 0.995 > 0.997 or a 1 > 0.003 > 0.991 > 0.997 (ignore the actual numbers, just to give an idea).
nevpugh308 said:
Oh, and what unit of measurement are we talking about ? 1.0 what ?
As has been mentioned, our real world isn't digital... a logical value of "1" usually means any value you like that is significantly different (by an arbitary amount) from an arbitarily decided baseline.
Imagine, if you will, a hard disk based on coloured ink, where the values "1" and "0" are represented by "red" and "yellow". Perhaps the hardware is such that a little of the previous colour might bleed through into the top layer (i.e most recently written) - the logic of the drive will have to assume a cutoff point to decide at what point "orange" becomes "red". By looking at the intensity of the red spots, you might be able to deduce what the previous colours were...
So, in answer to your question "1 what?" - there is not an actual unit of measurement, just a decision as to whether we want to interpret the reading as a "1" or a "0".
pdV6 said:
So, in answer to your question "1 what?" - there is not an actual unit of measurement, just a decision as to whether we want to interpret the reading as a "1" or a "0".
From the disc drive side it's the value of volts generated in the head coil. That's got to be really tiny in todays drives.
Note: I was going to say it's surprising it works at all, but that's when my compu
ThatPhilBrettGuy said:
From the disc drive side it's the value of volts generated in the head coil.
Sure - I was responding specifically to "1.0 what?". Its not "1.0" of anything, just an arbitary decision that values above a certain threshold represent the value "1"...
ThatPhilBrettGuy said:
Note: I was going to say it's surprising it works at all, but that's when my compu
jimmyjimjim said:
In short, yes.
Which is why the US department of defense format standard is to overwrite the data 7 times.
Why do they bother. Why not simply crush them, put them on a big bonfire or drop all the discs in water/acid – and make sure they are ruined totally.
Surely they do not recycle these discs, as after a good few years use there MTBF must be getting close, most will be out dated by cheaper and bigger capacity models. And would they really want to give them away to schools, colleges, etc…?
Rather than spending time re-formatting 7 times, just let them have an encounter with a BIG hammer.
jj
Well, personally, if I was to thrown away a hard drive, I'd physically destroy the platters.
In reality, what I, and most techies do, is take them out and use them as coasters, frisbee's, something to throw at next door's cat, etc, etc.
In the case of the DoD (paranoid), they format them seven times....then physically destroy them. Tax payers money in action. Better than our lot, who don't have to destroy disks, because they let them get stolen first.
In reality, what I, and most techies do, is take them out and use them as coasters, frisbee's, something to throw at next door's cat, etc, etc.
In the case of the DoD (paranoid), they format them seven times....then physically destroy them. Tax payers money in action. Better than our lot, who don't have to destroy disks, because they let them get stolen first.
when you say, the bit goes .95, .97, .99, 1
is not entirely true, its the other way around ! first time, then 0.97 , 0.93 etc as the media will degrade with time. As long as it exceeds a threshold the read head will treat it as a 1.
If you write a random set of ones and zero's in different lengths to the same sectors, it becomes more difficult, but not impossible to recover the data as, there will be several sets of signatures and they will become more difficult to determine, which ones and zeros belong to what.
is not entirely true, its the other way around ! first time, then 0.97 , 0.93 etc as the media will degrade with time. As long as it exceeds a threshold the read head will treat it as a 1.
If you write a random set of ones and zero's in different lengths to the same sectors, it becomes more difficult, but not impossible to recover the data as, there will be several sets of signatures and they will become more difficult to determine, which ones and zeros belong to what.
Gassing Station | Computers, Gadgets & Stuff | Top of Page | What's New | My Stuff