XPWRITE data integrity

Jim.Coomes at seagate.com Jim.Coomes at seagate.com
Wed Dec 17 15:25:42 PST 2003

* From the T10 Reflector (t10 at t10.org), posted by:
* Jim.Coomes at seagate.com

I would argue that the "Robust function" breaks the link between the data
block and the protection information. When the XOR product of the data
drives is calculated and written to the parity drive, new protection
information would have to be generated. When data for a failed data drive
is regenerated, the protection information (expected LBA/ref tag and some
knowledge of the app tag) has to be generated based on the resulting
regenerated user data.

With the "Simple function", the protection information is "untouched by
human hands". Although the protection information cannot be checked on the
transfer to or from the parity drive, the original protection information,
ref tag, guard  and app tag, is regenerated by the same XOR product used to
generate the user data.

I support the "Simple function". For a target implementation using "Simple
function", processing of the protection information is not different from
the user data The write data stream with protection information is XORed
with the associated fields on the media and the result is written back to
the media.


----- Forwarded by Jim Coomes/Seagate on 12/16/03 05:06 PM -----
                      "Symons, Timothy"                                                                                                
                      <Timothy_Symons at A        To:       "'Jim.Coomes at seagate.com'" <Jim.Coomes at seagate.com>                           
                      DAPTEC.com>              cc:                                                                                     
                      No Phone Info            Subject:  RE: XPWRITE data integrity                                                    
                      12/16/03 11:00 AM                                                                                                

I recommend that the "Robust function" should be adopted.
My interpretation of data integrity is that it is treated as part of the
data block throughout the data life, and any attempt to separate it would
compromise integrity.
The problem with the "simple function" is that it partitions data integrity
information from the body of the data.

             Tim Symons.
Phone: 719-884-8704
Email: Tim_Symons at adaptec.com

* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at t10.org

More information about the T10 mailing list