🚀 go-pugleaf

In April of this year, I purchased the "Enhanced HSC Format Utility",
part number QA-926AB-H7, for $100 because we have many Winchester
disk drives (RA82s) in a classified environment.  What prompted me
to buy it was the little blurb that stated, "This enhanced version
provides the functionality to overwrite data in all user sectors
(including revectored sectors) with user-specified patterns acceptable
to the Department of Defense."  (As the name of the product implies,
the disks must be on an HSC (except the HSC50).)

For those unfamiliar with classified work, whenever a Winchester
disk is replaced, the platters must be removed and destroyed if one
cannot overwrite every sector on the disk.  In itself, this would
be no big deal but DEC charges us $2800 for RA82s that are destroyed
since they cannot refurbish them.  Satisfying the Defense Investigative
Services (DIS) with this formatter would save my company lots of money
because we could declassify the disks and give them back to DEC.  The
fact that we still have many ancient RA82s which require FCOs to avoid
failures added some urgency to the purchase.

The regional DIS office requires that the procedure of writing a
particular pattern in every block (except those in the Format Control
Table) on the disk followed by random block dumps be performed three
times.  This requirement should be easily satisfied by using the new
FORMAT program followed by the DKUTIL program.

So we called DIS in for a demonstration of the product and we ran it.
(It's tough to get disks just to play with so there could be no "dress
rehearsals".)  We received a successful completion status but the
results were not what we expected or what was documented.  It seems
that the formatter will write all zeros to bad blocks rather than the
specified pattern that we selected.  It doesn't matter that the data
is gone (we couldn't prove it, that block may have always contained
zeros) -- what matters is that the pattern isn't what we specified.
The demo was a failure.

Talks with the SDI Controller Software Supervisor indicate that this
is a feature and the formatter does conform with the 1989 Industrial
Security Manual.  Unfortunately, this manual has been superseded by
the Government Declassification Specification of 1990.

It seems to me that to change the formatter to write the specified
pattern rather than zeros should be straightforward and easy but I
had the impression that nothing would be done.  The fact that it's
advertised to "overwrite data in *all* user sectors (including
revectored sectors) with user-specified patterns" indicates to me
that it isn't doing what it's supposed to do.

I might understand this stance if we were in a unique situation.  But
there must be hundreds of Digital customers who do classified work who
have RA82s going bad.  And each one must shell out $2800 each and every
time that happens.  I don't get it (except that it's a way of gouging
the customer since you can get RA82s from Midwest for about $1500).

Would SPRing it do any good?  How do other companies deal with
declassifying Winchesters?
-- 
Joe Barone
m1b@rayssd.ssd.ray.com
Gentlemen!  You can't fight in here!  This is the War Room!