SAM definition of SCSI Device

Ralph O. Weber ralphoweber at csi.com
Mon May 24 11:49:27 PDT 1999


* From the T10 Reflector (t10 at symbios.com), posted by:
* "Ralph O. Weber" <ralphoweber at csi.com>
*
Gerry,

I'm having a lot of trouble with the use of 'whispy' in the following:

} I see the the definition of "SCSI device" in SAM is a lot more wispy 
} than my concept of device, but I use the the term "SCSI device" as a
} fully functional SCSI entity that encompasses a target/initiator and
} logical unit (minimum) and optionally may include multiple LUs or
} multiple "ports".

SAM-2 says the following (which SAM said equally in object notation):

> ... an SCSI Device is composed of a Service Delivery Ports combined
> with an Initiator, or a Target, or both an Initiator and a Target.

> A Target is composed of a Target Identifier, a Task Manager, and one 
> or more Logical Units.

I don't see where SAM/SAM-2 says that a SCSI Device is anything less
than what you think it is.  The only thing remotely whispy about the
SAM definition is the number of words and pictures between the two
statements that form the totality of your definition.

The only other difference is that, as things currently stand, an SCSI
Device can have one and only one port.  That's a difference between
the current revision of SAM-2 and SAM (which does allow multiple
ports per SCSI Device).  The change comes directly from requests
|from the SCSI Working Group.

My understanding is that several members of the working group feel
strongly that an initiator must view an SCSI Device as having exactly
one port.  The initiator's view of a TBD SCSI Thing that has two ports
is that it (the initiator) is seeing two SCSI Devices.  Stated
differently, an initiator that scans a SCSI bus and finds two responding
target IDs cannot tell (without using SCSI commands) whether the
two IDs are shared by a single TBD SCSI Thing or belong to two
separate TBD SCSI Things.  By tradition, however, the two target
IDs are said to belong to two SCSI Devices (with no implied
relationship between the two).

Hopefully, the above discussion shows some of the thinking behind
the need for a new name.  Gene Milligan, for one, strongly disliked
multi-port SCSI Device.  Specific reasons were never given.  However,
there is something recursive about a multi-port SCSI Device containing
multiple SCSI Devices.

Keep the discussion coming.

Ralph...




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





More information about the T10 mailing list