Fwd: SAM-5: copy manager seldom an application client
Joe.Breher at hgst.com
Thu Jul 31 13:10:09 PDT 2014
Formatted message: <a href="http://www.t10.org/cgi-bin/ac.pl?t=r&f=r1407312_f.htm">HTML-formatted message</a>
Oops - I sent the below to John, rather than the reflector.
From: Joe Breher <joe.breher at hgst.com>
Subject: Re: SAM-5: copy manager seldom an application client
Date: July 31, 2014 1:53:26 PM MDT
To: John Lohmeyer <lohmeyer at t10.org<mailto:lohmeyer at t10.org>>
Certainly, there have been devices made, which incorporate a Copy Manager
derived from both the Device Server and the Application Client. Indeed, I
implemented such a device -- which had a non-negligible market presence --
back when EXTENDED COPY was a new thing. Of course, that was in a previous
life, and I've not integrated a copy manager in a device in some time.
I suppose T10 could entertain a proposal to make the change as suggested. I'd
need to see a more persuasive argument however.
Note that spc3r37 Figure 17 enumerates _examples_ of copy manager
configurations. As such, so inference should be made that the figure contains
an exhaustive catalog of all possible copy manager configurations.
More germane, however, is that Figure 17 illustrates a number of _system_
configurations - not _device_ configurations. Note that the various dashed
and dotted lines are transports. As such, the 'things' interconnected by
these transports are probably best thought of as separate devices.
Accordingly, the labels given to each of these 'things' -- of Application
Client, Device Server, and Copy Manager -- reflect more the roles that each
of these devices are playing in the illustrated configuration, rather than
objects in relation to each other within a given device.
This e-mail may contain confidential or legally privileged information of
HGST. If you are not the intended recipient, please notify us immediately by
responding to this e-mail and then deleting it from your system.
On Jul 31, 2014, at 12:49 PM, John Lohmeyer wrote:
* From the T10 Reflector (t10 at t10.org<mailto:t10 at t10.org>), posted by:
* John Lohmeyer <lohmeyer at t10.org<mailto:lohmeyer at t10.org>>
Douglas Gilbert <dgilbert at interlog.com> has
requested that I forward his below message to the T10 Reflector. Please cc
him on any response as he is not subscribed to the reflector.
-------- Forwarded Message --------
Subject: SAM-5: copy manager seldom an application client
Date: Wed, 30 Jul 2014 18:44:57 -0400
From: Douglas Gilbert <dgilbert at interlog.com>
Reply-To: dgilbert at interlog.com
To: T10 Reflector (t10 at t10.org<mailto:t10 at t10.org>)
<t10 at t10.org<mailto:t10 at t10.org>>
In the SCSI Classes overview (clause 4.6.1 on page 32 in
sam5r17.pdf) the UML diagram shows multiple inheritance
for the copy manager. It is saying that a copy manager
is (a type of) device server and is (a type of) application
client. The first half of that statement is true but
the second is only true for some standalone copy managers
(see spc4r37a.pdf clause 5.16.2, figure 17).
IMO the inheritance arrow (blue) from the copy_manager
to the application_client should be dropped. Other options
are to introduce a standalone_copy_manager which the
copy_manager contains 0 or 1 of. Another level of indirection
is needed to identify those standalone_copy_managers that use
T10 techniques to talk to the device servers owning the source
and destination of the copy (as the standalone_copy_manager
can use other techniques such as a network). Then the UML
diagram can say the t10_standalone_copy_manager is (a type
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to
majordomo at t10.org<mailto:majordomo at t10.org>
More information about the T10