Jonathan Wright schrieb: > Alexander Skwar wrote: >>>A good option for more speed would be to use aespipe to encrypt >>>/dev/zero and write that output to the disk. >> >> Why do that? Overwritten data is already pretty much >> irrecoverable. Or do you know of ONE instance, where >> those rescue companies restored an OVERWRITTEN (ie. >> not something simple as burned or such) drive? >> >> And if you fear, that the CIA or FBI might recover >> data - use a metal shredder... > > Actually, even if you format a hard drive, it's still relatively easy to > get the data off.
Well, but that's only so, because a format normally doesn't rewrite the whole device but only some "header". > I can't remember the name of the program to do it, I bet it's based on "Secure Deletion of Data from Magnetic and Solid-State Memory" by Peter Gutmann (see <http://www.cs.auckland.ac.nz/~pgut001/pubs/secure_del.html>). > but if you want to > securely erase a hard drive according to NSA/CIA/FBI standards, There's no such thing. Those agencies recommend to incinerate harddrives containing sensitive data. > it needs > 37 passes using RANDOM data! No, it doesn't. That used to be true *AGES* ago with RLL/MFM drives, but no more with current drives. Nowadays, thanks to the improved reliability of harddrives, a simple overwrite with 0's is good enough. Back in May 2003, the german computermagazine c't send drives to ontrack, vogon and ibas with overwritten files. In *no* cases those companies could recover data. Even if data was just overwritten with 0! Those companies are only sucessful, when recovering data from burned or otherwise physically destroyed/inaccessible drives. Or do you have *PROVE* that those companies actually CAN recover overwritten data? I would *REALLY* be interested! Alexander Skwar -- gentoo-user@gentoo.org mailing list
