RFC: Sanatization Standards for possible Eraser inclusion


I was wondering if anyone could help me. I have trying to come up with a list
of the various country/government data sanitizations but apart from the DoD ones
I am having a hard time getting any detailed specifics on how they are performed.

The main reason I am bringing this up is for inclusion in a future releases of
eraser. I'm aware of the arguments that there's only a need for so many, and I
would have agreed with this my self. However thinking about it. The more standardized
wiping methods that eraser supports the more useful it is to people who have to
adhere to these standards.

Would love to hear what people think of this, especially the Heidi guys.

So far I am aware of the following:

+ US DoD 5220.22-M.
http://www.dss.mil/isp/odaa/documents/n ... 6-5220.pdf
3 to 7 passes

+ US Navy standards NAVSO P-5239-26. (MFM)
One Pass 0xffffffff, One Pass 0xbfffffff, one Pass Random, Drive verified after wiping

+ US Navy standards NAVSO P-5239-26. (RFF)
One Pass 0xffffffff, One Pass 0x27ffffff, one Pass Random, Drive verified after wiping

+ US Air Force, AFSSI5020
One pass 0x00, One pass 0xff, One pass random. 10% of drive read to verify wiping.

+ US Army AR 380-19

+ RCMP Hard Drive Secure Information Removal and Destruction Guidelines

+ British HMG Infosec Standard No.5.

+ German VSItR Standard.
7 passes with specific passes, Drive verified after wiping

+ Australian ASCI 33.

+ Russian GOST R 50739-95.
1 pass random.

+ North Atlantic Treaty Organization - NATO standard
Alternating 0x00 and 0xff for 6 passes followed by one pass random.

Non Stnadardized.
+ Peter Gutmann's algorithm.
http://www.cs.auckland.ac.nz/~pgut001/p ... e_del.html
4 random passes, 27 specific passes, 4 random passes

+ Bruce Schneier's algorithm.
2 pass specific, 5 passes using cryptographically secure pseudo-random sequence

edit 20070712-1413 added Canada
edit 20070718-1413 added US Army
Speaking of this topic, I wonder why nobody has done the following:

  • Created sets of specific, unique files on a system (known data) of various types/formats (e.g. binaries, images, movies, archives, etc.),
  • Erased these specific file sets with each of these several methods available in Eraser,
  • Sent in the disk drive to a well-known recovery service for recovery.

And then published the results as to how successful was the recovery/erasing with respect to each erasure method as well as the recovery method used (or level of service performed).

It would be interesting to see if there is a method of erasure which IS recoverable through these disk recovery services, or if even One Pass 0xFFFFFFFF is sufficient (i.e. fastest possible), and that multiple passes are simply redundant and provide no measurable benefit.

I wonder how much of the basis for these methods are based upon old technologies (e.g. MFM, RLL) and lore from the late 1980's.
mike808 said:
  • Sent in the disk drive to a well-known recovery service for recovery.

Interesting idea, but I don't think you'll find anyone willing to pay the several hundred euro/dollar/poud price tage these services request when recovering files.
I wouldn't be surprised if someone proposed a grant to measure the effectiveness of a freely available GPL'd software. Perhaps a corporation with a financial interest in complying with government or internal standards/requirements for data destruction would fund thesis work on the topic.

Peter Gutmann's paper, on which Eraser (and its competitors) claims are based is over ten years old now. The largest disk drives available at the time were on the order of single-digit GBs. We're closing in on two orders of magnitude larger drives in 2007.

How relevant are the technologies available from the recovery industry perspective that were available in 1996 (when Peter did his research for the paper) when applied to 2007 technology hard drives?

Has the recovery technology kept up with disk drive technology?

That's the sort of question that the recovery industry itself (or governments with an interest in their agencies keeping digital media secure) might be interested in funding such a study or thesis proposal.