scrubbing tapes in hp-ux

[g3jza]

Hi,
How do you scrub tapes (LTO in my case) that contain sensitive data? I've checked 'mediainit' , but this command should only handle the scrubbing of disks. Does anyone know about something similar to this? The tapes are going to be scrubbed physically, but I want to make sure. Thanks

[Viktor Balogh]

Hi,

one good way is to overwrite data with dd:

# dd if=/dev/zero of=

Regards,
Viktor

[Steven Schweda]

> How do you scrub tapes [...]

Define "scrub".

> [...] that contain sensitive data? [...]

Define "sensitive".

> [...] The tapes are going to be scrubbed
> physically, [...]

Define "scrubbed physically".

> [...] I want to make sure.

Define "sure".

If you _really_ want the data gone, then fire
(or at least a very high temperature) works
very well. If money is no object, then data
may be recoverable from a tape which has been
"erased" by many other methods.

If you want the data gone for most practical
purposes, then a big magnet can do a
reasonable job in a reasonable amount of
time, and does not destroy the tape. Writing
over a whole tape (using "dd", or some other
program) also does a reasonable job, but is
much slower.

A Forum or more general Web search should
find many previous related discussions
involving disks and/or tapes. Many people
_say_ that they want perfect security (data
destruction), but many of those are not
willing to destroy the medium itself, and
that destruction is the only perfectly secure
method.

[Doug Burton]

You can use a third party that will destroy the tapes for you and give you a document certifying the destruction.

A good way to get rid of those old pesky tapes and meet your legal obligations as well.

[g3jza]

Steven:
By scrubbing I mean, making it hard for someone to possibly recover it.
Sensitive in our case = mostly database data.
The physical scrubbing (destroying the tapes) is not done by us but by some other company. I just wanted to 'pre-srub' the data on those tapes so that it will not be possible for that company(which is doing the physical destruction) to read those data :) .

Thanks for all your ideas.

[Viktor Balogh]

> making it hard for someone to possibly recover it.

making it hard not equals to making it impossible, but I know what you mean ;)

[Larry Klasmier]

I would assume you have a non-disclosure agreement with the third party company who will destroy the tapes, correct? In our case that company is also the company that handles the offsite storage of our backup tapes, guessing this probably the same for you?

Seems to me you are just making extra work for yourself.

[Bill Hassell]

> big magnet...

For DDS, DLT and LTO, this not only destroys the data, it will destroy the tape completely by erasing the pre-recorded servo information. The tape can never be reused as the drive will reject it.

Technically, (for modern media) you can simply write a single record (a few bytes) then rewind the tape. The drive will write an end-of-data marker and now the tape cannot be positioned to read anything beyond this marker. For use inside your company, this is by far the easiest and fastest method. Try it on one of your tapes and see if anyone can read the data past the first record.

[Steven Schweda]

> > big magnet...
>
> For DDS, DLT and LTO, this not only
> destroys the data, it will destroy the tape
> completely by erasing the pre-recorded
> servo information. The tape can never be
> reused as the drive will reject it.

Really? I'm pretty sure that I've
bulk-erased DAT and DLT tapes, and I remain
unaware of any "servo information" on these
media. LTO is too modern for me, so I know
nothing about those, but I'm dubious. Can
you offer a credible reference for this
claim? (Or is it confined to the oral
tradition?)

[Steven Schweda]

> For DDS, DLT and LTO, [...]

A quick Google search for:

magnetic tape servo track

found, among others:

http://www.webuyusedtape.net/degaussing.html

I quote:

[...] Media products such as 3480, 3490e,
DLT, SDLT can be degaussed and the media
reused. However, 3590, LTO, 9840 and 9940
cartridges cannot be degaussed without
destroying the servo track(s). [...]

These folks apparently do this for a living,
so they may know something. (So, that's one
out of three right in that "DDS, DLT and LTO"
list, as I suspected.)

> Seems to me you are just making extra work
> for yourself.

I agree, but if your plan is to have the
media destroyed, and not reused, then the big
magnet still seems to me to be a reasonable
tool for the job. (If you insist on doing
the job.) If the plan is to reuse the media,
and LTO tape does seem to use a servo track,
then bulk erasure would seem to be a big
mistake. Unless you have the equipment (and
time) to re-write the servo track (on each
medium).

[VK2COT]

Hello,

HP-UX 11.31 has a DoD 5220.22-M compliant tool to run disk scrubbing. Hp-UX is the only
Unix that I know of to have such a native
OS tool. To check if your version
supports it:

# mediainit | grep scrub

Of course, in some very sensitive environments,
as you said, media has to be destroyed
physically in the end...

Cheers,

VK2COT

[Bill Hassell]

> DDS, DLT, SDLT...no servos

Correct. My blanket statement left off the magnet bias effects which can render virtually all tapes as unusable. In the good old days, the bulk eraser was called a degausser and would run off AC power thus creating constantly changing magnetic polarities. You would turn on the unit, move it close to the media, then slowly pull it away so as not to leave a magnetic bias. This bias decreases the signal to noise ratio which becomes increasingly more critical as the tape density increases. A permanent magnet will definitely cause magnetic bias unless the user constantly twists the magnet to simulate alternating magnetic polarity. This can affect even the ancient 1/2inch reel to reel tapes (first hand experience).

[Steven Schweda]

> My blanket statement left off the magnet
> bias effects which can render virtually all
> tapes as unusable. [...]

This sounds to me like (still) more nonsense.

First, I never said which kind of "big
magnet" to use, so I (intentionally) did not
exclude the use of a big, AC electromagnet.

Second, its not at all clear to me why
magnetizing a whole tape uniformly (even if
that were possible using a big permanent
magnet) would render it unusable, any more
than magnetizing selected parts of it (by,
say, writing data to it normally) would. But
perhaps there's something fundamental about
magnetic tape recording which I don't
understand. (Or at least one of us doesn't.)

> [...] (first hand experience).

My first-hand experience differs from yours.

I sure do enjoy the folklore, but, as before,
if you could offer a credible reference for
this claim, too, then it might be more
plausible. (But I suspect that this one's
about as good as the previous one.)

[Bill Hassell]

http://www.dw-world.de/staticfiles/pdf/dwfz/Magnetic_Sound_Recording.pdf

Section 4.5.2 (and others) explain the effects of DC bias, caused by permanent magnet fields. Recording heads cannot create a strong enough field to overcome this residual magnetism.

[Steven Schweda]

When did we change the discussion topic to
Magnetic _Sound_ Recording? When dealing
with zeros and ones, rather than audio
waveforms, concepts like distortion and noise
lose much utility.

Is this what worried you?:

DC magnetization on a recorded tape
cannot be removed and increases the
noise level permanently.

What that actually means is that any DC bias
in an audio recording can't be removed
>>without erasing the recorded audio<<. It
does _not_ mean that the tape can never be
demagnetized.

> [...] Recording heads cannot create a
> strong enough field to overcome this
> residual magnetism.

Says who? I believe that you're mistaken.
(He said, trying to avoid prohibited personal
abuse.) An _audio_ recording head may not
normally be driven hard enough to saturate
the tape, because one normally tries to keep
an audio signal within the (roughly) linear
range of the medium. In digital recording
(as with an audio erase head) this kind of
limit does not apply.

That document appears to contain considerable
valuable information, but pulling some
irrelevant warning out of its middle does not
constitute a wise use of a potentially
valuable resource. But if you find any
reliable information which is also relevant,
then please let us know. (Something about
recording digital data on a magnetic medium,
for example.)