Question about ROD token "well-known" type

Joseph Glider gliderj at almaden.ibm.com
Mon Apr 25 07:20:20 PDT 2011


Formatted message: <a href="http://www.t10.org/cgi-bin/ac.pl?t=r&f=r1104252_f.htm">HTML-formatted message</a>

Thanks Fred.
Given what you say, I wouldn't want to require that storage systems that 
want to support the main copy offload function to also be required to 
support this almost-WRITE-SAME capability. It increases the development 
burden for storage systems to do so and it would be a shame to delay 
support of the main function based on needing to support a function that 
isn't needed by initiators because they already have a tool to get the job 
done.
 Is support of well-known tokens separately configurable already, or can 
it be?
Jody Glider
STSM, Storage Systems and Servers Research
Tieline: 457-1853
External: 408-927-1853
From:
"Knight, Frederick" <Frederick.Knight at netapp.com>
To:
"Joseph Glider" <gliderj at almaden.ibm.com>, <t10 at t10.org>
Date:
04/25/2011 04:46 AM
Subject:
RE: Question about ROD token "well-known" type
Sent by:
owner-t10 at t10.org
Yes, similar to WRITE SAME, but not identical.
The primary difference is the ability to use multiple LBA extents (a 
scatter write of zeros vs. a single extent in WRITE SAME).   The 
similarity to the token operations, is that the device knows the data 
(zeros in this case), and the application has a reference to that data 
(the token – the well known zero token in this case).
		Fred
From: Joseph Glider [mailto:gliderj at almaden.ibm.com] 
Sent: Thursday, April 21, 2011 5:58 PM
To: t10 at t10.org
Subject: RE: Question about ROD token "well-known" type
This sounds like WRITE SAME, how is it different or better? It doesn't 
seem to be solving a problem similar to the one that proxy read/write was 
proposed for, or am I missing the point? 
Thanks, 
Jody Glider
STSM, Storage Systems and Servers Research
Tieline: 457-1853
External: 408-927-1853 
From: 
<david.black at emc.com> 
To: 
<gliderj at almaden.ibm.com>, <t10 at t10.org> 
Date: 
04/21/2011 02:43 PM 
Subject: 
RE: Question about ROD token "well-known" type 
Sent by: 
owner-t10 at t10.org
There will be a single well-known ROD token type in the next version of 
11-080 and 11-059, and the name of this will be changed to reflect exactly 
what it is.  The purpose of that predefined ROD token will be to serve as 
a source of zeroes for the WRITE USING TOKEN command in 11-059 - to be 
precise, when the WRITE USING TOKEN command presents that predefined ROD 
token as a source of data, the written user data is all 0’s and the 
written protection information, if any, is all F’s. 
Thanks,
--David 
From: owner-t10 at t10.org [mailto:owner-t10 at t10.org] On Behalf Of Joseph 
Glider
Sent: Thursday, April 21, 2011 5:14 PM
To: t10 at t10.org
Subject: Question about ROD token "well-known" type 
I am wondering about the 'well-known ROD token type' as mentioned on page 
4 of the 11-080 proposal. There is no description of how this is used, 
except to say this: Editor’s Note: First two bytes of the type currently 
indicate the token format, except for FF which is for exceptions. For the 
well-known SBC-3 tokens, see 11-059. 
But I can't find any mention of this in the 11-059 doc. 
Can somebody point me in the right direction? 
Jody Glider
STSM, Storage Systems and Servers Research
Tieline: 457-1853
External: 408-927-1853 



More information about the T10 mailing list