Wiping Old Laptop

securityguy

But actually reading the random device will be a lot quicker than anything else

I was surprised, actually, at how slow /dev/urandom is on Linux. OSX manages about 10MB/sec (both on a recent Air and a rather older Mini). A Linux VM running on the same hardware gets between 6MB/sec and 8MB/sec, depending on kernel (3.X seems to be faster than 2.6.X). Neither's fast enough to keep a disk busy, even via USB.

A Solaris VM, again on the same hardware, gets 60MB/sec (in each case "dd bs=64k count=1024 if=/dev/urandom of=/dev/null").

Usual caveats about consulting someone who knows what they're doing if you really need security, but encrypting /dev/zero with a random key is almost certainly good enough for disk sanitisation, and is interestingly about ten times faster on Linux.

ubuntu@ubuntu-VirtualBox:~$ dd if=/dev/zero  bs=64k count=1024 2>/dev/null | openssl enc -aes-128-ctr -md sha1 -k "$(dd if=/dev/random bs=32 count=1 2>/dev/null)" | dd bs=64k of=/dev/null iflag=fullblock
1024+1 records in
1024+1 records out
67108880 bytes (67 MB) copied, 0.857505 s, 78.3 MB/s
ubuntu@ubuntu-VirtualBox:~$ dd if=/dev/urandom bs=64k count=1024 of=/dev/null
1024+0 records in
1024+0 records out
67108864 bytes (67 MB) copied, 7.66481 s, 8.8 MB/s
ubuntu@ubuntu-VirtualBox:~$ 

tronator

S0litaire wrote: »

For data wiping you should do more than one pass of ANY software using random (or as random as possible) data.

Do you have a link of any case where somebody successfully recovered data from a completely overwritten hard disk? Most of these claims are of theoretically nature and require a lot of time and/or manpower. Do you really think somebody will spend years and years to recover some files from John Doe's hard disk?

http://en.wikipedia.org/wiki/Gutmann_method
Quote:
The National Bureau of Economic Research criticized Gutmann's claim that intelligence agencies are likely to be able to read overwritten data, citing a lack of evidence for such claims.[3] Companies specializing in recovery of damaged media (e.g., media damaged by fire, water or otherwise) cannot recover completely overwritten files. No private data recovery company currently claims that it can reconstruct completely overwritten data.

prowla

On a VM with 2 vCPUs and 16 GB RAM:

$ dd bs=64k count=1024 if=/dev/urandom of=/dev/null
1024+0 records in
1024+0 records out
67108864 bytes (67 MB) copied, 13.5956 s, 4.9 MB/s
$

On a HP BL460c G6 blade with 1x 6-core 2-thread Xeon X5650 2.67 GHz & 18 GB RAM:

$ dd bs=64k count=1024 if=/dev/urandom of=/dev/null
1024+0 records in
1024+0 records out
67108864 bytes (67 MB) copied, 7.39878 seconds, 9.1 MB/s
$

Not exactly blisteringly fast.

And for file copy (this server has SAN storage)...

$ dd bs=64k count=1024 if=/dev/urandom of=blah
1024+0 records in
1024+0 records out
67108864 bytes (67 MB) copied, 7.45672 seconds, 9.0 MB/s
$ 

$ ls -l blah
-rw------- 1 xxxx xxxx 67108864 Feb 10 14:04 blah
$ 

$ time cp blah blah2

real    0m0.091s
user    0m0.005s
sys    0m0.082s
$

So, the file copy wins hands-down.

securityguy

tronator wrote: »

Do you have a link of any case where somebody successfully recovered data from a completely overwritten hard disk? Most of these claims are of theoretically nature

Or for modern disks, not even theoretical.

wakeupalarm

How do you securely wipe a SSD?

securityguy

wakeupalarm wrote: »

How do you securely wipe a SSD?

It depends on how nervous you are.

If you write zeros over it, you will erase somewhere around 90% of its capacity. The problem is that the wear-levelling mechanism keeps a proportion of the space out of circulation (perhaps 8GB on a 128GB device) and uses those blocks to replace blocks that have been repeatedly written to. Unlike with a hard drive, where bad block management is the responsibility of the operating system's file system code, and therefore stamping zeros over the disk will attempt to write to good and bad blocks alike, on an SSD the block retirement is done by the disk controller and therefore you cannot access the retired blocks via the SATA interface. Hence if you scribble zeros over the entirety of an SSD which has been used hard, there will be retired blocks which you will not be able to overwrite, but which may still contain sensitive information. It will total at most, say, 10% of the capacity, and it may be a problem for you.

The ATA "Secure Erase" operation is supposed to work on SSDs and overwrite the retired blocks as well, but (again) if you're very nervous, it's a matter of some conjecture how you'd verify that, given you can't read the retired blocks anyway.

This is why best practice is to encrypt SSDs as soon as you start using them. That way, the blocks that get retired are encrypted, and when you come to dispose of the disk, you just overwrite the volume key and those retired blocks are not recoverable. In fact, assuming modern processors, this is best practice anyway, on magnetic as well as solid-state storage: the cost of full-disk encryption is very low on processors like Intel i5 and above which have the AES round function in hardware.

However, if you were feeling _very_ nervous, there's always the risk that the volume key is in a block that is retired, in which case the encrypted volume is only protected by the passphrase that you use to boot the machine. But if you're that nervous, you'd physically destroy the SSD anyway.

50Twuncle

monkey84 wrote: »

Hello,


Here's hoping there is somebody out there who is more computer savvy than I and can help! :-)


I want to wipe my old Toshiba laptop before sending it to be recycled. I know that I could just reformat the machine but am aware that this doesn't completely remove everything. I have also heard of programs like Killdisk and Nuke that do seem to do this.


My question is, is it necessary for me both to reformat and run one of these programs as well or would just running program get rid of everything?
Which should I do first if I do need to do both? Does anyone know of any programs that are better than the above?


The CD drive on my old machine doesn't work so I can't create any sort of recovery or boot disk, but I'm not sure this is necessary as I'm not planning reusing the laptop anyway.


Many thanks in advance.

Returning to the original question
Since you don't want to reuse the laptop and I assume are not part-ex'ing it - why not just remove the HDD and drill a few holes through the platter ?

rmg1

Or, as has been suggested before, put the HDD into a caddy and keep it as a spare/backup device?
You could probably even get a USB-powered caddy to use it as a portable drive.

prowla

If re-using the hard disk for something else, it would be a good idea to keep its resultant empty caddy with the laptop.

S0litaire

tronator wrote: »

Do you have a link of any case where somebody successfully recovered data from a completely overwritten hard disk? Most of these claims are of theoretically nature and require a lot of time and/or manpower. Do you really think somebody will spend years and years to recover some files from John Doe's hard disk?

At the moment, No I don't have any link to any case where data has been recovered. That is if the software doing the wipe does what it said.

The point is "Do you always trust a piece of software to do what it says?" Most security people will say if you can't review the code and compile it yourself, assume it's compromised in some fashion.

I don't 100% trust programs like "dd" even though others have reviewed the code. I don't have the knowledge or skills to review the code for myself but I can compile programs and I rely on a degree of trust in other professionals.

You also have to remember that "urandom" and "random" are "Psudo-Random" data streams, so it's possible (however unlikely) to recreate identical outputs from "urandom" and "random".

So running the wipe multiple times is just good practice. (their is a method to the madness of the DoD specs requiring multiple wipes.)

It was also though a few years ago to be theoretically impossible to recover data from RAM chips after the power was removed. But today all you need is a Linux PC with a free RAM slot and a can of "Compressed air" or a Thermos of "Liquid nitrogen" and anyone can cool a RAM chip so that any data stored on it can be read for 10's of mins or even an hour or so after the power to the chip has been removed.

My point is!
What's theoretical "today" might be possible in a well equipped lab "tomorrow" and then "the day after" to anyone with the ability to follow simple instructions.

So getting into good habits now saves a lot of hassle in the future.