SAM-x and initiator identifiers

Jim Hafner hafner at almaden.ibm.com
Mon Apr 2 12:36:27 PDT 2001


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

Is the following a correct statement?

"The initiator port identifier implicitly contains the relative target port
identifier."

Here's why I think this is the case:
In many cases, the language of the specs say that certain state information
(like reservations, task sets, unit attentions, mode pages) etc. are
handled with respect to the "initiator identifier".   However, whatever
defines that initiator identifier must also include something that
identifies the target port through which an initiator is seen from the
viewpoint of the logical unit.  That is required to disambiguate the
implied SCSI domain on which an initiator lives.

Examples:
1) SPI -- here, the common interpretation of the initiator identifier is
the SCSI bus address.  However, that is insufficient to uniquely identify
an initiator on one bus (through one target port) and another initiator on
another bus (through a second port).  Hence, the "initiator identifier"
must implicitly contain the relative target port.
2) FCP -- here, the common interpretation of the initiator identifier is
the FC N_PortID (3byte S_ID).  However, as for SPI, this is also
insufficient to uniquely identify an initiator connected on one target port
|from an initiator on another because the two target ports could be on
different FC fabrics where N_PortIDs could be duplicated.

Is there another interpretation?
Jim Hafner

*
* 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