Name that multi-port SCSI device

George Ericson gericson at mindspring.com
Mon Jun 21 21:01:11 PDT 1999


* From the T10 Reflector (t10 at symbios.com), posted by:
* "George Ericson" <gericson at mindspring.com>
*
Brian,
Sorry to take so long getting back to you.
Responses to your questions inline.
George

-----Original Message-----
From: owner-t10 at Symbios.COM [mailto:owner-t10 at Symbios.COM]On Behalf Of
briandm at excite.com
Sent: Wednesday, June 02, 1999 4:48 PM
To: GEricson at BAnet.net; briandm at excite.com; '-T10 Reflector'
Cc: roweber at acm.org
Subject: Re: RE: Name that multi-port SCSI device


* From the T10 Reflector (t10 at symbios.com), posted by:
* briandm at excite.com
*
George,

I think we're there.
Thanks for your patience.
You're right.  I did not see before that X and Xa were not the same logical
unit.  (Does the "a" in "Xa" have anything to do with "Port a"?)
----------------
GME: Yes
----------------

I like your model better than mine.
(BTW The target in my model is not defined by physical hardware but by
shared Logical Units -- all LUs shared between (among?) ports belong to the
same target, regardless of which LUs are attached to which port).
----------------
GME: This will work.
----------------

I do have an issue with this statement from your last note:
On Sat, 29 May 1999 11:19:24 -0400, George M. Ericson wrote:
> ...
> So, one can stay within the boudaries of SAM-2.8 and yet still describe
all
> of the complex relationships between ports and LUs on a controller that
have
> been asked for.
> ...
I don't think the relationship between X and Xa was described in SAM-2.8 or
any other standard.
----------------
GME: True.  However, the concept of stacking devices is.  See the
discussions on 8-byte LUNs and on Virtual Devices.
----------------

Do we now need two new names, one for the multi-ported target Tab(X,Y)
and one for relationship between LU Xa and LU X associated with different
targets?
----------------
GME: The first is discussed in sections 4.7, 4.7.1, and 4.7.2.  No
particular name is given.
However, I am in favor of using the term Multi-Ported SCSI Device when the
existance of more than one port is important to the context.  Single-Ported
SCSI Device when the existance of exactly one port is important to the
context.  When the number of ports is not important to the context, the term
SCSI Device is sufficient.
The relationship between LU Xa and LU X is vendor implementation specific.
I don't see a need (yet) for the model to define a special name.  (BTW, RAID
subsystems use this type of relationship, only one to many.)
----------------
Aside from the names, here's what I'd like to see addressed:
(1) I would like to have the option to treat Tab(X,Y) as either a single
target: Tab(X,Y) or two targets: Ta(X,Y) and Tb(Xa, Ya) (e.g. have a
TARGET_RESET on Ta not interrupt I/O on Tb).  (I think this is the issue
Gene referred to).
----------------
GME: The R2.8 model allows you to do both.
----------------

(2) For two targets Ta(X,Y) and Tb(Xa,Z) I'd like to be able to have LUs X
and Xa look like the same LU --- exactly the same as LU X in Tab(X).
----------------
GME: The R2.8 model allows you to do this.  Implementation would have to
make sure that Xa reflected X accurately.  Note here that a reset to Ta
should affect Ta(X), Tb(Xa), and Ta(Y), but not Tb(Z).  This because the
task manager of Ta simply forwards the target reset to Ta as an LU Reset to
X and Y.  The LU resets are processed by the device servers of X and Y
across all initiators, regardless of port.  In your implementation, Xa acts
as an initiator to Xa and should propagate the Unit Attentions that result.
----------------
Brian


Brian McKean                             Voice: (303)381-4246
Mylex Corporation External Products      Fax:   (303)413-0464
4900 Pearl East Circle Suite 104         brian at mylexboulder.com
Boulder, CO 80301-6108
http://www.mylex.com





_______________________________________________________
Get your free, private email at http://mail.excite.com/
*
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at symbios.com

*
* 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