Combined initiator & target names in SAM-2

Jim Hafner hafner at almaden.ibm.com
Fri Nov 9 12:19:27 PST 2001


* From the T10 Reflector (t10 at t10.org), posted by:
* "Jim Hafner" <hafner at almaden.ibm.com>
*

Peter,

Your comment is noted. However, there are different ways around this issue
(and they could have been applied to the port name question as well).

Namely, we have to ask ourselves what "device" is being named?   Is it the
physical device or the virtual device providing a specific function.  In
your "may be completely unaware of each other" statement, one can interpret
that as being two independent devices that happen to share the same piece
of hardware.  In that case, you have one target device with its name and
one initiator device and its name.

On the other hand, suppose we allow for two different names (one that names
the initiator function and one that names the target function).  Have we
lost or gained anything?  The fact that it has two different names doesn't
add any features over and above a picture of two virtual devices each with
a unique name.  They will have to be managed separately in any case.  And
how can you tell the difference between the two cases?

So, the net is that I don't really think it matters in practice one way or
another.  I would agree to either option: go back to separate names or to
stay with the decision of the WG (which I guess you weren't at during this
discussion).

I think the sense of the WG was that there was no compelling reason to have
separate names so why make the model more complex.

Jim Hafner


Peter Johansson <PJohansson at acm.org>@t10.org on 11/09/2001 04:49:57 AM

Sent by:  owner-t10 at t10.org


To:   NCITS T10 <T10 at t10.org>
cc:   SBP-3 <SBP3 at ISG.Apple.com>
Subject:  Combined initiator & target names in SAM-2



* From the T10 Reflector (t10 at t10.org), posted by:
* Peter Johansson <PJohansson at ACM.org>
*
 From the minutes of the CAP working group, agenda item 4.6:

"In a related issue the initiator and target device names in a SCSI
target/initiator device will be combined to form a single name, subject to
the one per protocol rule as well."

This does not work for SBP; I also suspect that it is, on general
principal, ill-advised since initiator and target functionality are
orthogonal to each other. In representative software stack implementations
for PCs, initiator and target code may be completely unaware of each other.

I suggest that separate initiator and target names continue to be
maintained, as they have in the past. A subsequent excerpt from the CAP
minutes, reproduced below, seems to indicate ample precedent for
maintaining separate names.

"George also asked if the target/initiator device initiator port name and
target port name can be combined in a single name. Jim Hafner responded by
noting that both iSCSI and SRP have different name definitions for
initiator and target ports, resulting in the need for separate, different
names in the architecture model."


Regards,

Peter Johansson

Congruent Software, Inc.
98 Colorado Avenue
Berkeley, CA  94707

(510) 527-3926
(510) 527-3856 FAX

PJohansson at ACM.org

*
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at t10.org



*
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at t10.org




More information about the T10 mailing list