Perfectly clear to me. When you erase a disk, ... you erase it. What
else would you do? :-) 

Precisely why our storage management group does whatever it is that they
do to decommission disks and then contracts with the disk manufacturers
to destroy them. 

Regards, 
Richard Schuh 


-----Original Message-----
From: The IBM z/VM Operating System [mailto:[EMAIL PROTECTED] On
Behalf Of Alan Altmark
Sent: Wednesday, January 31, 2007 2:23 PM
To: [email protected]
Subject: Re: ICKDSF Release 16

On Wednesday, 01/31/2007 at 03:57 CST, "McKown, John" 
<[EMAIL PROTECTED]> wrote:
> Have you looked at SAE? It has a stand-alone ERASE capability. You can
> even IPL it off of the DVD in the HMC! Or DASD or Tape.

I find such products ... interesting.  If you look at the S/390 Command 
Reference for ESS (SHARK), you will find that the ERASE CCW or
formatting 
record 0 of a track are documented to "erase" data.  But when you read 
what it says about erasure, things get fuzzy.  To wit:

"A record is said to be erased when it is rendered incapable of being
read 
by the
commands and facilities normally used for reading recorded data from the

device.
The manner in which the erase operation is performed is overwriting the 
record
referenced with pad characters and then erasing the remainder of the 
track."

Of course, it uses the word "erase" in the definition of the operation,
so 
who knows what it will do.  Consider, too, that modern drives separate
the 
appearance of ECKD from reality.  It can simply mark some metadata that 
says "erased" which the microcode will respect.  But take the RAID
drives 
out and read them elsewhere, and who knows what you will find.  Then 
there's that nasty word "normally used" in there.  It makes me nervous.

And if you have a moving cursor algorithm, rewriting the "same" record
may 
not, in fact, rewrite the same record.  It may get written and then
index 
pointers are updated.

I find lots of information about "data security erase" for tapes, but
not 
for disks.

Methinks I must contact a Reverened Engineer to find out what gives
these 
days...

Alan Altmark
z/VM Development
IBM Endicott

Reply via email to