-------- Original Message --------
Subject: Comments on: CbCS 'correction' proposals
Date: Thu, 6 Mar 2008 17:25:53 -0500
From: Sivan Tal <SIVANT@il.ibm.com>
To: Ralph Weber <roweber@ieee.org>
CC: owner-t10@t10.org, "'t10@t10.org'" <t10@t10.org>, Kevin D Butt <kdbutt@us.ibm.com>, "David Black" <Black_David@emc.com>

* From the T10 Reflector (t10@t10.org), posted by:
* Sivan Tal <SIVANT@il.ibm.com>
*
<snip>

Comment 1:
Thanks for correcting the SA usage. However, one of the "features" didn't
make it to the correct usage. That is the requirement that the creation of
the SA had included an authentication step.
Now, since the minimum SA parameters do not include the information
required to determine whether the authentication step had been skipped or
not, this involves maintaining additional info that is not specified in the
standard. While this can still be done, I suspect a better way to require
authentication is to make a change to the IKEv2-SCSI part as follows:
If the selected USAGE_TYPE SA parameter is "CbCS authentication and
credential encryption" then the authentication step must not be skipped (in
other words, SA_AUTH_NONE must not be selected).

<gigantic snip>
  
Hopefully, the following new proposal addresses this issue.

http://www.t10.org/ftp/t10/document.08/08-138r0.pdf

However, I have not had time to confirm the suitability of the
proposal with the SA creation gurus (i.e., things may get a
little dicey when CAP reviews the plan).

All the best,

.Ralph