Conservation DistList Archives [Date] [Subject] [Author] [SEARCH]

Subject: HD-ROM


From: Robert J. Milevski <milevski>
Date: Monday, July 17, 1995
This message just came in today.
It is a follow-up to the other one I sent earlier on this subject.

The following appeared on Archives and is reproduced here without
the knowledge or consent of the authors

    Date: 17 Jul 95
    From: Marc Wolfe <marc.wolfe [at] arch2__nara__gov>
    Subject: Ion beam milling--detailed report

    I prepared this for the records management listserve and thought
    ARCHIVES might be interested also.  Apologies to any of you who
    subscribe to both lists and got two copies.

    This is not a commercial product yet, but based on the
    discussion I just had with Roger Stutz, one of the scientists
    responsible for developing the technology, IMHO it's got the
    potential to be a real a winner.

    The scientists have considered the implications of
    software/hardware dependency, drawing on the body of knowledge
    that RMs and archivists work with.

    Stutz recommended a January 1995 Scientific American article
    written by Rotenberg (sp?) on ensuring the longevity of
    electronic data on which the LANL guys based their documentation
    strategy.  I haven't read the article but plan to.

    Stutz gives the maximum density for binary data storage at 23K
    megabits per square inch.  The density can be varied at will.
    The kicker is that information can be stored in different
    formats on the *same* data device. This includes *human*
    readable (i.e., very, very small images) or pixel format as well
    as binary data.  Needless to say the 23k megabit density is for
    binary data.  The 23K is currently the maximum density, but
    Stutz said they expect incremental improvements.

    LANL has experimented with their own in house data to determine
    appropriate densities for the different storage methods.  Images
    are stored in varying densities as microforms based on density
    of the original; a photograph of an object is stored as a much
    higher density than the image of a document.

    The documentation standard for reading the non-microform data is
    stored in microform if appropriate, or in ASCII/EBCDIC.  This
    appears to be the key to platform independence.  Any
    documentation necessary to read binary/digital data can be
    stored in microform due to the information density of the

    The cost of materials (assuming stainless steel) is $20 per
    terabyte (one thousand gigabytes, I believe).  This compares to
    $13K for magtape and $2500 for current optical disk media.  Cost
    to write the data, including labor (but I'm pretty sure that
    doesn't include pre-processing arrangement of records) is
    "pennies" per page (less than  10 cents, with most documents at
    around 3 cents).

    The medium doesn't have to be in the form of a pin.  In theory
    the technique could involve "microengraving" on any surface.  In
    practice a regularly shaped medium is best.  Tested forms
    include stainless steel tape and the equivalent of 3x5" cards.

    Currently the material is being advanced as ROM, which in theory
    makes it most useful  as an archival or long-term retention
    medium. However, once a master is made, provided disposition
    instructions are included in the documentation, the master can
    be recopied without the information which has become due for
    disposal.  [job tenure, folks!]

    I suspect that as the technology matures it will be made
    suitable for the types of storage we currently use hard drives,

    The steel tape form of the medium would appear especially
    suitable (IMHO) for unattended recopying, or remastering,
    because the recopying process has more writeable square footage
    in the least amount of apparent space.

    Stutz said that with minimal care the steel pin will last 1000
    years and with very vigilant care 5000.  Minimal care involves
    monitoring the environment for humidity, temperature flux, and
    acidity. Almost any physical substance can be used.  Rubidium,
    mentioned in the press release, is rated as physically stable
    for this use as 10,000 years.

    The LANL group has done their homework on the RM side of things.
    Everything necessary to read the binary data is there in
    human-readable format, which was my major concern.  It's up to
    us to determine how long our employers will need to read the
    data, and to make provisions for the timely disposition of data
    (be it transfer to the National Archives in my case, to destroy
    after 20 years for example).

    One final word about human readability.  For microforms we've
    used the example of a light source and a magnifying lens as the
    base level of technology to keep microfilm human readable.  It
    appear that the entry-level just got a little higher.  Granted,
    you need a relatively sophisticated piece of equipment (atomic
    force *micro*scope), but if the data is being stored digitally
    on magtape/CD ROM, you still need a computer to read it.  If a
    civilization can build a computer it can build an atomic force
    microscope (neither machine is demonstrably more complex to
    manufacture.)  The maximization of the storage density
    demonstrated by HD-ROM allow the information necessary to read
    the binary data to be stored in a software independent format.

    marc.wolfe [at] arch2__nara__gov

    PS:  This is not a commercial endorsement.  I don't have any
    direct financial interest in the process, or any link to
    inventors other than my conversation with Roger Stutz.  As a
    private individual I *will* definitely be interested in the IPO

                  Conservation DistList Instance 9:12
                   Distributed: Friday, July 21, 1995
                        Message Id: cdl-9-12-004
Received on Monday, 17 July, 1995

[Search all CoOL documents]