On Sat, May 18, 2013 at 3:34 AM, Dan Egli <[email protected]> wrote:
> Ok that makes sense. What about software raid? Would software raid still be > one you wouldn't "shred" a file on? Or would it operate just the same as a > regular independent drive for secure erasing? > > Software raid may still have failed a drive and added a hot spare--but this you can at least control completely. > > > And while it's true that recovery programs would be stymied by just writing > zeros, a physical hardware recovery is still possible in certain lab > situations. No. Simply not true. This is fantastical rumor that spread around the industry. Writing zeroes is sufficient. The way data is encoded as it is written on the platter is highly complex--even the drive vendors would have a nigh-unto impossible time recovering any data if possible at all. When a stream of zeroes is written, it isn't just zeros that hit the platter--it is encoded (eg 8b10b--likely much more complicated) then one or more layers of crc codes are added to ensure the data can be read back in the event of errors in the platter medium. These rumors hinge on the idea of the possibility of being able to read the previous signal by, for example, shifting the heads off of track center and hoping there's enough residual to reconstruct it. That might have worked on floppy disks back in the day, but is simply not going to work with the high density platters we have today--let alone perpendicular recording methods now used. /* PLUG: http://plug.org, #utah on irc.freenode.net Unsubscribe: http://plug.org/mailman/options/plug Don't fear the penguin. */
