In a message dated 6/29/2005 9:24:24 A.M. Central Daylight Time,  
[EMAIL PROTECTED] writes:

Just out of  curiosity, why multiple passes.  I know that they must have 
their  reasons...  Something like residual data on the edges of the track 
or  some such, since there is a possibility of the track not being 
recorded in  the exact location of the previous write?




The precise location of recorded bits around the track moves around  
somewhat.  That is why the controller does something called "read with  offset" 
when 
re-reading a block that produced a data check (aka "head  shaking").  The 
read-write mechanism/transducer is moved a tiny distance to  the left of the 
center 
of where the track is supposed to be, the data is read,  if still bad then 
the transducer is moved a tiny distance to the right of the  center...  If the 
data cannot be read correctly (i.e., no data check  indication) and you have 
moved the transducer so far to either side that it is  now within the tolerance 
for the adjacent track, then you give up on reading  that block.
 
Another issue is residual magnetism.  When you magnetize a substance,  a 
large % of the molecules are aligned in a certain way.  When you erase  that 
data, 
not all the molecules get realigned back to their original  direction.  This 
is why skilled recording experts with expensive, sensitive  equipment can 
recover recorded data that was over-written or erased.  The  cost of such 
recovery 
is high.  The cost of erasing data to the point that  it cannot be recovered 
by an expert is also high.  The trade-off is how  valuable is the data to you 
and what would it cost you if someone else got the  data.
 
Here are the DoD's standards:  
_http://72.14.207.104/search?q=cache:dRHWf5Fg2LQJ:security.ouhsc.edu/docs/DoD_5220.doc+DoD+5220&hl=en_
 
(http://72.14.207.104/search?q=cache:dRHWf5Fg2LQJ:security.ouhsc.edu/docs/DoD_5220.doc+Do
D+5220&hl=en) 
 
On the line that says non-removable rigid disk it says that in order to  
clear the data all you need to do is to overwrite all addressable locations 
with  
a single character, such as writing a full track of X'00'.  But in order to  
"sanitize" the disk you must degauss the disk, destroy the disk, or use process 
 "d", which means to Overwrite all addressable  locations with a character, 
its complement, then a random character and  verify.  This would require at 
least 4 revolutions of the track, the first  3 of which are writing and the 4th 
is reading, then some CPU time to do the  verify.  So if it takes X hours to 
download the data and you want it  sanitized, it will take 4X hours to sanitize.
 
Bill  Fairchild

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html

Reply via email to