here's the comment area for today's blog article found at
http://zabkat.com/blog/30Aug09-destroy-files.htm
blog: secure file/folder deletion
Moderators: fgagnon, nikos, Site Mods
-
- Site Admin
- Posts: 16310
- Joined: 2002 Feb 07, 15:57
- Location: UK
-
- Bronze Member
- Posts: 110
- Joined: 2003 Jun 16, 18:54
it would seem logical in a case of distributed writes as you describe for solid state memory that the shred function would zero the file with the most recent changes as mapped by up-to-date file allocation information, not writing zeroes somewhere else.
it is surely from previous writes in the file's history that file data allocated elsewhere might remain, and therein lies the risk.
it is surely from previous writes in the file's history that file data allocated elsewhere might remain, and therein lies the risk.
-
- Site Admin
- Posts: 16310
- Joined: 2002 Feb 07, 15:57
- Location: UK
-
- Bronze Member
- Posts: 94
- Joined: 2008 Jan 15, 10:08
For solid-state devices with write distribution it's not the filesystem the distributes the writes, but the device firmware - which means the information isn't just invisible to top-level applications, your OS doesn't (and can't) know about it either.
I haven't heard about any USB flash drives that does this, though - and a lot of SATA SSDs don't even have the feature either, because it means somewhat more expensive design.
PS: before anybody starts yelling that single-pass zero overwrite is unsafe, read the Gutmann article that Nikos links to... even Gutmann himself says that these days, he doubts that (even with the batshit insane expensive lab equipment) you'd be able to recover data after a single-pass overwrite.
I haven't heard about any USB flash drives that does this, though - and a lot of SATA SSDs don't even have the feature either, because it means somewhat more expensive design.
PS: before anybody starts yelling that single-pass zero overwrite is unsafe, read the Gutmann article that Nikos links to... even Gutmann himself says that these days, he doubts that (even with the batshit insane expensive lab equipment) you'd be able to recover data after a single-pass overwrite.