Minutes of banded command study group
gerry.houlder at seagate.com
Fri Apr 20 13:17:45 PDT 2012
Formatted message: <a href="http://www.t10.org/cgi-bin/ac.pl?t=r&f=r1204202_f.htm">HTML-formatted message</a>
The banded commands study group was held by telecon/ webex on Friday April
20, starting 11:00am CDT.
Frank Wadjija (Dell)
David Black (EMC)
Dan Colegrove (HGST)
Curtis Ballard (HP)
Fred Knight (NetApp)
Mike Fitzpatrick (Toshiba)
Marty Czekalski (Seagate)
Tim Feldman (Seagate)
Gerry Houlder (Seagate)
Jim Hatfield (Seagate)
Steve Finch (WD)
Jim Malina (WD)
Curtis Stevens (WD)
(missed one attendee that joined late)
Discussion started with 12-034r1 -- banded command concepts. These opinions
were clearly stated:
- The device should be a new peripheral device type in standard Inquiry,
not type 00h with additional bit in extended Inquiry.
- The group was unsure about the "device performs the equivalent of .."
descriptions. There was support for the need to explain why certain
behaviors cause sub-optimal performance but there is also a thought that
the standard should just say "there will be sub-optimal performance"
without saying why. There are words in the Copy Manager section of SPC-4
that does some similar explanation, maybe words like that will work.
- For write error recovery, there may be some statements in SSC that are
- Use of UNMAP to force write pointer to change was considered
undesirable. Need to create a new command for this. Main reason is UNMAP
has certain expectation for direct access device type but behavior for the
sequential write bands in new device type is different and behavior in
random write bands has to be the same as direct access device type.
- Use of GET LBA STATUS the retrieve write pointer is also not desired,
this needs to be a new command.
- A write command that goes past the end of a band should end in CHECK
CONDITION status, even if the next band is unwritten.
Discussion on 12-010r1 -- MDC project proposal, pointed out the need to
revise the dates and some changes to the description. Rev. 2 with these
changes will be posted to T10 web site early next week.
The telecon concluded at 12:45pm.
More information about the T10