Subject: RE: transfer ready not on block boundaries?
Date: Fri, 4 May 2007 16:07:06 -0700
From: "Robert Snively" <rsnively@brocade.com>
To: <Bob.Nixon@emulex.com>, <t10@t10.org>
X-Message-Number: 7752
Formatted message: HTML-formatted message

Our engineering staff indicate that it is not rare for devices to
perform transfer operations that do not align with block boundaries.
To our knowledge, most devices have already elected to deal with
this characteristic, whether it is pretty or not.
At this point, I believe that interoperability problems may occur
if block alignment is made a requirement.
Bob Snively
________________________________
From: owner-t10@t10.org [mailto:owner-t10@t10.org] On Behalf Of
Bob.Nixon@emulex.com
Sent: Tuesday, May 01, 2007 1:47 PM
To: t10@t10.org
Subject: SBC: transfer ready not on block boundaries?
As far as I have been able to tell, our collection of standards permit a
transfer ready to be issued by a block device for an amount of data not
a multiple of the block size. This causes serious extra complexity in
the design of devices that support concurrent use of DIF and any kind of
cryptographic security.
Are there members of this distribution that know that non-block-bounded
transfer ready is being done? (This is only a check for existence. Do
not identify the product if you prefer not to.)
Are there members of this distribution willing to describe a noteworthy
advantage to doing it? How about for doing it efficiently?
Thanks in advance!
   - bob nixon, Emulex