SPC-3 device name supported identifier types
George Penokie
gop at us.ibm.com
Fri Dec 17 08:54:52 PST 2004
* From the T10 Reflector (t10 at t10.org), posted by:
* George Penokie <gop at us.ibm.com>
*
This is a multipart message in MIME format.
--=_alternative 005CFDDD86256F6D_=
Content-Type: text/plain; charset="US-ASCII"
Paul,
Are you sure you are referring to target devices? Almost everything in
SCSI deals with target ports not target devices. A target device as
defined by SAM-3 contains:
a) Zero or more target device names;
b) One or more SCSI target ports, each containing:
A) A task router;
B) A target port identifier;
C) An optional relative port identifier; and
D) An optional target port name;
and
c) One or more logical units.
Bye for now,
George Penokie
Dept 2C6 114-2 N212
E-Mail: gop at us.ibm.com
Internal: 553-5208
External: 507-253-5208 FAX: 507-253-2880
Paul von Behren <Paul.Vonbehren at Sun.COM>
12/15/2004 10:31 AM
To
George Penokie/Rochester/IBM at IBMUS
cc
"Elliott, Robert (Server Storage)" <elliott at hp.com>, owner-t10 at t10.org,
t10 at t10.org
Subject
Re: SPC-3 device name supported identifier types
George Penokie wrote:
>
> Rob,
>
> This is one strange mess. First things first, the shall in the
statement
> 'The Device Identification VPD page shall have one or more
> identification descriptors for the SCSI target device.' was not
> intended. It was only supposed to be a shall when the target device
> contained well-known logical units. For commands sent to target
devices
> that do not have well known logical units it was supposed to be a
> should. I would change the wording to:
>
> _If the SCSI target device contains a well known logical unit_ the
> Device Identification VPD page shall have one or more identification
> descriptors for the SCSI target device. _If the SCSI target device
does
> not contain a well known logical unit the Device Identification VPD
page
> should have one or more identification descriptors for the SCSI target
> device. _Each SCSI target device identification descriptor_, if any,_
> shall have the ASSOCIATION field set to 2h (i.e., SCSI target device)
> and the IDENTIFIER TYPE field set to:
> a) 2h (i.e., EUI-64-based);
> b) 3h (i.e., NAA); or
> c) 8h (i.e., SCSI name string).
> The Device Identification VPD page shall contain identification
> descriptors_, if any,_ for all the SCSI target device names for all
the
> SCSI transport protocols supported by the SCSI target device.
>
> Comments!!!!!!!!!!!!
Would it be possible to get a stronger requirement for multi-LU devices?
I can understand that both target and LU identifiers for a simple disk
or tape drive may be overkill. But the target ID is a huge benefit for
management of RAID/virtualization devices. I fear for my life when
I tell SMI-S client application developers that Array target device
IDs are optional :-)
Here's the second sentence of the second paragraph above:
"If the SCSI target device does not contain a well known logical unit
the Device Identification VPD page should have one or more
identification descriptors for the SCSI target device. "
How about following it with:
"If the SCSI target device does not contain a well known logical unit
and supports more than one logical unit, the Device Identification VPD
page shall have one or more identification descriptors for the SCSI
target device. "
or something equivalent and less wordy...
thanks, Paul
>
> Bye for now,
> George Penokie
>
> Dept 2C6 114-2 N212
> E-Mail: gop at us.ibm.com
> Internal: 553-5208
> External: 507-253-5208 FAX: 507-253-2880
>
>
>
>
> *"Elliott, Robert (Server Storage)" <elliott at hp.com>*
> Sent by: owner-t10 at t10.org
>
> 12/14/2004 05:34 PM
>
>
> To
> <t10 at t10.org>
> cc
>
> Subject
> RE: SPC-3 device name supported identifier types
>
>
>
>
>
>
>
>
> Although you say "There is no requirement that a target device have a
> name", Kevin Marks pointed out in the last SAT call that SPC-3 rev 21b
> section 7.6.4.11.1 includes this paragraph:
> "The Device Identification VPD page shall have one or more
> identification descriptors for the SCSI target device. Each SCSI
target
> device identification descriptor shall have the ASSOCIATION field set
to
> 10b (i.e., SCSI target device) and the IDENTIFIER TYPE field set to:
2h,
> 3h, or 8h."
>
> This was added by 02-419r7, which worded it (page 12) as: "A SCSI
target
> device shall have one or more identification descriptors. Each SCSI
> target device identification descriptor shall contain the ASSOCIATION
> field set to 2h (i.e., SCSI target device) and the IDENTIFIER TYPE
field
> set to: 2h, 3h, or 8h."
>
> An alternative solution would be to remove that requirement; then SAT
> would simply not provide one if it has to conjure one itself (i.e., if
> it is not in a bridge that already has one available). When SAM-3
> discusses target device names (sam3r14 section 4.7.2) it discusses
"zero
> or more", not "one or more". Which rule did you intend in 02-419r7,
the
> SAM-3 rule or the SPC-3 rule?
> --
> Rob Elliott, elliott at hp.com
> Hewlett-Packard Industry Standard Server Storage Advanced Technology _
> __https://ecardfile.com/id/RobElliott_
>
>
>
>
------------------------------------------------------------------------
> *From:* George Penokie [mailto:gop at us.ibm.com] *
> Sent:* Tuesday, December 14, 2004 2:54 PM*
> To:* Elliott, Robert (Server Storage)*
> Cc:* Ralph O. Weber; t10 at t10.org*
> Subject:* RE: SPC-3 device name supported identifier types
>
>
> Rob,
>
> Reset, I cross-connected my logical units and devices. But the point
> still stands. On logical units we have to deal with history which is
the
> only reason we allow type 1 and it does cause no end of trouble. For
> target devices and target ports there was no history on names so we
> wisely eliminated type 1 from them. I see no reason to take a giant
step
> backward. There is no requirement that a target device have a name so
if
> there is no way SATA can do, so what.
>
> Bye for now,
> George Penokie
>
> Dept 2C6 114-2 N212
> E-Mail: gop at us.ibm.com
> Internal: 553-5208
> External: 507-253-5208 FAX: 507-253-2880
>
>
>
> *"Elliott, Robert (Server Storage)" <elliott at hp.com>*
>
> 12/14/2004 02:29 PM
>
>
> To
> George Penokie/Rochester/IBM at IBMUS, "Ralph O. Weber"
<roweber at ieee.org>
> cc
>
> Subject
> RE: SPC-3 device name supported identifier types
>
>
>
>
>
>
>
>
>
>
> Type 1h names are already allowed for logical units.
>
> They'll be new for device names. Since you wanted device names to be
> subject to the whims of the underlying protocols, this is the ATA
> protocol's whim :-)
>
> --
> Rob Elliott, elliott at hp.com
> Hewlett-Packard Industry Standard Server Storage Advanced Technology _
> __https://ecardfile.com/id/RobElliott_
>
>
>
>
------------------------------------------------------------------------
> *From:* George Penokie [mailto:gop at us.ibm.com] *
> Sent:* Tuesday, December 14, 2004 2:19 PM*
> To:* Ralph O. Weber*
> Cc:* Elliott, Robert (Server Storage)*
> Subject:* Re: SPC-3 device name supported identifier types
>
>
> Rlaph,
>
> Yes, I have been watching. It's only an issue if the solution effects
> non-SATA things. Which allowing type 1 names for logical units would.
>
> Bye for now,
> George Penokie
>
> Dept 2C6 114-2 N212
> E-Mail: gop at us.ibm.com
> Internal: 553-5208
> External: 507-253-5208 FAX: 507-253-2880
>
>
> *"Ralph O. Weber" <roweber at ieee.org>*
>
> 12/14/2004 12:03 PM
>
>
> To
> George Penokie/Rochester/IBM at IBMUS
> cc
> "Elliott, Robert (Server Storage)" <elliott at hp.com>
> Subject
> Re: SPC-3 device name supported identifier types
>
>
>
>
>
>
>
>
>
>
>
>
> George,
>
> In that case, you need to start paying much closer
> attention to the discussions of 04-218r? in SAT and CAP.
> For a couple of go-rounds now, they have been telling
> you that ATA devices cannot be trusted to provide
> enough information to make WWIDs for logical units.
>
> The only question has been exactly what form the non-
> unique names will take.
>
> All the best,
>
> .Ralph
>
> George Penokie wrote:
>
> >
> > Rob,
> >
> > I have a problem with this. It gives an opening to allow the use of
> > non-unique WWIDs for logical units. The type 1h ("T10 vendor
> > identification") format cannot be guaranteed to be unique, that is
the
> > reason we require one of the other types for logical units. System
> > management cannot take a change on this getting screwed up it's too
> > important.
> >
> > Bye for now,
> > George Penokie
> >
> > Dept 2C6 114-2 N212
> > E-Mail: gop at us.ibm.com
> > Internal: 553-5208
> > External: 507-253-5208 FAX: 507-253-2880
> >
> >
> >
> >
> > "Elliott, Robert (Server Storage)" <elliott at hp.com>
> > Sent by: owner-t10 at t10.org
> >
> > 12/02/2004 04:58 PM
> >
> >
> > To
> > <t10 at t10.org>
> > cc
> >
> > Subject
> > SPC-3 device name supported identifier types
> >
> >
> >
> >
> >
> >
> >
> >
> >
> > * From the T10 Reflector (t10 at t10.org), posted by:
> > * "Elliott, Robert (Server Storage)" <elliott at hp.com>
> > *
> > During the SAT teleconference today, Kevin Marks pointed out that
SPC-3
> > 7.6.4.11.1 requires that each logical unit report a device name
using
> > one of these identifier types: 2h (EUI-64-based), 3h (NAA), or 8h
(SCSI
> > name string). (that rule was introduced by 02-419, starting with
> > 02-419r4)
> >
> > SAT can do this, but needs to use the type 1h ("T10 vendor
> > identification") format in some environments because it is not
going to
> > have the tools to create a 2h, 3h, or 8h identifier (see 04-218r6).
> >
> > So, I propose (as a letter ballot comment) that SPC-3 add type 1h
to the
> > list, and include a "should be 2h, 3h, or 8h" rule as is done in
> > 7.6.4.11.3 for logical unit names.
> >
> > --
> > Rob Elliott, elliott at hp.com
> > Hewlett-Packard Industry Standard Server Storage Advanced
Technology
> > https://ecardfile.com/id/RobElliott
> >
> >
> > *
> > * For T10 Reflector information, send a message with
> > * 'info t10' (no quotes) in the message body to majordomo at t10.org
> >
>
>
>
>
--=_alternative 005CFDDD86256F6D_=
Content-Type: text/html; charset="US-ASCII"
<br><font size=2 face="sans-serif">Paul,</font>
<br>
<br><font size=2 face="sans-serif">Are you sure you are referring to target
devices? Almost everything in SCSI deals with target ports not target
devices. A target device as defined by SAM-3 contains:</font>
<br>
<br><font size=2 face="sans-serif">a) Zero or more target device names;</font>
<br><font size=2 face="sans-serif">b) One or more SCSI target ports, each
containing:</font>
<br><font size=2 face="sans-serif">A) A task router;</font>
<br><font size=2 face="sans-serif">B) A target port identifier;</font>
<br><font size=2 face="sans-serif">C) An optional relative port identifier;
and</font>
<br><font size=2 face="sans-serif">D) An optional target port name;</font>
<br><font size=2 face="sans-serif">and</font>
<br><font size=2 face="sans-serif">c) One or more logical units.</font>
<br>
<br><font size=2 face="sans-serif"><br>
Bye for now,<br>
George Penokie<br>
<br>
Dept 2C6 114-2 N212<br>
E-Mail: gop at us.ibm.com<br>
Internal: 553-5208<br>
External: 507-253-5208 FAX: 507-253-2880<br>
<br>
</font>
<br>
<br>
<br>
<table width=100%>
<tr valign=top>
<td width=40%><font size=1 face="sans-serif"><b>Paul von Behren <Paul.Vonbehren at Sun.COM></b>
</font>
<p><font size=1 face="sans-serif">12/15/2004 10:31 AM</font>
<td width=59%>
<table width=100%>
<tr>
<td>
<div align=right><font size=1 face="sans-serif">To</font></div>
<td valign=top><font size=1 face="sans-serif">George Penokie/Rochester/IBM at IBMUS</font>
<tr>
<td>
<div align=right><font size=1 face="sans-serif">cc</font></div>
<td valign=top><font size=1 face="sans-serif">"Elliott, Robert (Server
Storage)" <elliott at hp.com>, owner-t10 at t10.org, t10 at t10.org</font>
<tr>
<td>
<div align=right><font size=1 face="sans-serif">Subject</font></div>
<td valign=top><font size=1 face="sans-serif">Re: SPC-3 device name supported
identifier types</font></table>
<br>
<table>
<tr valign=top>
<td>
<td></table>
<br></table>
<br>
<br>
<br><font size=2><tt><br>
<br>
George Penokie wrote:<br>
> <br>
> Rob,<br>
> <br>
> This is one strange mess. First things first, the shall in the statement
<br>
> 'The Device Identification VPD page shall have one or more <br>
> identification descriptors for the SCSI target device.' was
not <br>
> intended. It was only supposed to be a shall when the target device
<br>
> contained well-known logical units. For commands sent to target devices
<br>
> that do not have well known logical units it was supposed to be a
<br>
> should. I would change the wording to:<br>
> <br>
> _If the SCSI target device contains a well known logical unit_ the
<br>
> Device Identification VPD page shall have one or more identification
<br>
> descriptors for the SCSI target device. _If the SCSI target device
does <br>
> not contain a well known logical unit the Device Identification VPD
page <br>
> should have one or more identification descriptors for the SCSI target
<br>
> device. _Each SCSI target device identification descriptor_, if any,_
<br>
> shall have the ASSOCIATION field set to 2h (i.e., SCSI target device)<br>
> and the IDENTIFIER TYPE field set to:<br>
> a) 2h (i.e., EUI-64-based);<br>
> b) 3h (i.e., NAA); or<br>
> c) 8h (i.e., SCSI name string).<br>
> The Device Identification VPD page shall contain identification <br>
> descriptors_, if any,_ for all the SCSI target device names for all
the <br>
> SCSI transport protocols supported by the SCSI target device.<br>
> <br>
> Comments!!!!!!!!!!!!<br>
<br>
Would it be possible to get a stronger requirement for multi-LU devices?<br>
I can understand that both target and LU identifiers for a simple disk<br>
or tape drive may be overkill. But the target ID is a huge benefit
for<br>
management of RAID/virtualization devices. I fear for my life when<br>
I tell SMI-S client application developers that Array target device<br>
IDs are optional :-)<br>
<br>
Here's the second sentence of the second paragraph above:<br>
"If the SCSI target device does not contain a well known logical unit
<br>
the Device Identification VPD page should have one or more <br>
identification descriptors for the SCSI target device. "<br>
<br>
How about following it with:<br>
"If the SCSI target device does not contain a well known logical unit<br>
and supports more than one logical unit, the Device Identification VPD
<br>
page shall have one or more identification descriptors for the SCSI <br>
target device. "<br>
<br>
or something equivalent and less wordy...<br>
<br>
thanks, Paul<br>
<br>
> <br>
> Bye for now,<br>
> George Penokie<br>
> <br>
> Dept 2C6 114-2 N212<br>
> E-Mail: gop at us.ibm.com<br>
> Internal: 553-5208<br>
> External: 507-253-5208 FAX: 507-253-2880<br>
> <br>
> <br>
> <br>
> <br>
> *"Elliott, Robert (Server Storage)" <elliott at hp.com>*<br>
> Sent by: owner-t10 at t10.org<br>
> <br>
> 12/14/2004 05:34 PM<br>
> <br>
> <br>
> To<br>
> <t10 at t10.org><br>
> cc<br>
> <br>
> Subject<br>
> RE:
SPC-3 device name supported identifier types<br>
> <br>
> <br>
> <br>
> <br>
> <br>
> <br>
> <br>
> <br>
> Although you say "There is no requirement that a target device
have a <br>
> name", Kevin Marks pointed out in the last SAT call that SPC-3
rev 21b <br>
> section 7.6.4.11.1 includes this paragraph:<br>
> "The Device Identification VPD page shall have one or more <br>
> identification descriptors for the SCSI target device. Each SCSI target
<br>
> device identification descriptor shall have the ASSOCIATION field
set to <br>
> 10b (i.e., SCSI target device) and the IDENTIFIER TYPE field set to:
2h, <br>
> 3h, or 8h."<br>
> <br>
> This was added by 02-419r7, which worded it (page 12) as: "A
SCSI target <br>
> device shall have one or more identification descriptors. Each SCSI
<br>
> target device identification descriptor shall contain the ASSOCIATION
<br>
> field set to 2h (i.e., SCSI target device) and the IDENTIFIER TYPE
field <br>
> set to: 2h, 3h, or 8h."<br>
> <br>
> An alternative solution would be to remove that requirement; then
SAT <br>
> would simply not provide one if it has to conjure one itself (i.e.,
if <br>
> it is not in a bridge that already has one available). When SAM-3
<br>
> discusses target device names (sam3r14 section 4.7.2) it discusses
"zero <br>
> or more", not "one or more". Which rule did you
intend in 02-419r7, the <br>
> SAM-3 rule or the SPC-3 rule?<br>
> -- <br>
> Rob Elliott, elliott at hp.com<br>
> Hewlett-Packard Industry Standard Server Storage Advanced Technology
_<br>
> __https://ecardfile.com/id/RobElliott_<br>
> <br>
> <br>
> <br>
> ------------------------------------------------------------------------<br>
> *From:* George Penokie [mailto:gop at us.ibm.com] *<br>
> Sent:* Tuesday, December 14, 2004 2:54 PM*<br>
> To:* Elliott, Robert (Server Storage)*<br>
> Cc:* Ralph O. Weber; t10 at t10.org*<br>
> Subject:* RE: SPC-3 device name supported identifier types<br>
> <br>
> <br>
> Rob,<br>
> <br>
> Reset, I cross-connected my logical units and devices. But the point
<br>
> still stands. On logical units we have to deal with history which
is the <br>
> only reason we allow type 1 and it does cause no end of trouble. For
<br>
> target devices and target ports there was no history on names so we
<br>
> wisely eliminated type 1 from them. I see no reason to take a giant
step <br>
> backward. There is no requirement that a target device have a name
so if <br>
> there is no way SATA can do, so what.<br>
> <br>
> Bye for now,<br>
> George Penokie<br>
> <br>
> Dept 2C6 114-2 N212<br>
> E-Mail: gop at us.ibm.com<br>
> Internal: 553-5208<br>
> External: 507-253-5208 FAX: 507-253-2880<br>
> <br>
> <br>
> <br>
> *"Elliott, Robert (Server Storage)" <elliott at hp.com>*<br>
> <br>
> 12/14/2004 02:29 PM<br>
> <br>
> <br>
> To<br>
> George
Penokie/Rochester/IBM at IBMUS, "Ralph O. Weber" <roweber at ieee.org><br>
> cc<br>
> <br>
> Subject<br>
> RE:
SPC-3 device name supported identifier types<br>
> <br>
> <br>
> <br>
> <br>
> <br>
> <br>
> <br>
> <br>
> <br>
> <br>
> Type 1h names are already allowed for logical units.<br>
> <br>
> They'll be new for device names. Since you wanted device names
to be <br>
> subject to the whims of the underlying protocols, this is the ATA
<br>
> protocol's whim :-)<br>
> <br>
> -- <br>
> Rob Elliott, elliott at hp.com<br>
> Hewlett-Packard Industry Standard Server Storage Advanced Technology
_<br>
> __https://ecardfile.com/id/RobElliott_<br>
> <br>
> <br>
> <br>
> ------------------------------------------------------------------------<br>
> *From:* George Penokie [mailto:gop at us.ibm.com] *<br>
> Sent:* Tuesday, December 14, 2004 2:19 PM*<br>
> To:* Ralph O. Weber*<br>
> Cc:* Elliott, Robert (Server Storage)*<br>
> Subject:* Re: SPC-3 device name supported identifier types<br>
> <br>
> <br>
> Rlaph,<br>
> <br>
> Yes, I have been watching. It's only an issue if the solution effects
<br>
> non-SATA things. Which allowing type 1 names for logical units would.<br>
> <br>
> Bye for now,<br>
> George Penokie<br>
> <br>
> Dept 2C6 114-2 N212<br>
> E-Mail: gop at us.ibm.com<br>
> Internal: 553-5208<br>
> External: 507-253-5208 FAX: 507-253-2880<br>
> <br>
> <br>
> *"Ralph O. Weber" <roweber at ieee.org>*<br>
> <br>
> 12/14/2004 12:03 PM<br>
> <br>
> <br>
> To<br>
> George
Penokie/Rochester/IBM at IBMUS<br>
> cc<br>
> "Elliott,
Robert (Server Storage)" <elliott at hp.com><br>
> Subject<br>
> Re:
SPC-3 device name supported identifier types<br>
> <br>
> <br>
> <br>
> <br>
> <br>
> <br>
> <br>
> <br>
> <br>
> <br>
> <br>
> <br>
> George,<br>
> <br>
> In that case, you need to start paying much closer<br>
> attention to the discussions of 04-218r? in SAT and CAP.<br>
> For a couple of go-rounds now, they have been telling<br>
> you that ATA devices cannot be trusted to provide<br>
> enough information to make WWIDs for logical units.<br>
> <br>
> The only question has been exactly what form the non-<br>
> unique names will take.<br>
> <br>
> All the best,<br>
> <br>
> .Ralph<br>
> <br>
> George Penokie wrote:<br>
> <br>
> ><br>
> > Rob,<br>
> ><br>
> > I have a problem with this. It gives an opening to allow
the use of<br>
> > non-unique WWIDs for logical units. The type 1h ("T10
vendor<br>
> > identification") format cannot be guaranteed to be
unique, that is the<br>
> > reason we require one of the other types for logical units.
System<br>
> > management cannot take a change on this getting screwed
up it's too<br>
> > important.<br>
> ><br>
> > Bye for now,<br>
> > George Penokie<br>
> ><br>
> > Dept 2C6 114-2 N212<br>
> > E-Mail: gop at us.ibm.com<br>
> > Internal: 553-5208<br>
> > External: 507-253-5208 FAX: 507-253-2880<br>
> ><br>
> ><br>
> ><br>
> ><br>
> > "Elliott, Robert (Server Storage)" <elliott at hp.com><br>
> > Sent by: owner-t10 at t10.org<br>
> ><br>
> > 12/02/2004 04:58 PM<br>
> ><br>
> >
<br>
> > To<br>
> >
<t10 at t10.org><br>
> > cc<br>
> >
<br>
> > Subject<br>
> >
SPC-3 device name supported identifier types<br>
> ><br>
> ><br>
> ><br>
> >
<br>
> ><br>
> ><br>
> ><br>
> ><br>
> ><br>
> > * From the T10 Reflector (t10 at t10.org), posted by:<br>
> > * "Elliott, Robert (Server Storage)" <elliott at hp.com><br>
> > *<br>
> > During the SAT teleconference today, Kevin Marks pointed
out that SPC-3<br>
> > 7.6.4.11.1 requires that each logical unit report a device
name using<br>
> > one of these identifier types: 2h (EUI-64-based), 3h (NAA),
or 8h (SCSI<br>
> > name string). (that rule was introduced by 02-419,
starting with<br>
> > 02-419r4)<br>
> ><br>
> > SAT can do this, but needs to use the type 1h ("T10
vendor<br>
> > identification") format in some environments because
it is not going to<br>
> > have the tools to create a 2h, 3h, or 8h identifier (see
04-218r6).<br>
> ><br>
> > So, I propose (as a letter ballot comment) that SPC-3 add
type 1h to the<br>
> > list, and include a "should be 2h, 3h, or 8h"
rule as is done in<br>
> > 7.6.4.11.3 for logical unit names.<br>
> ><br>
> > --<br>
> > Rob Elliott, elliott at hp.com<br>
> > Hewlett-Packard Industry Standard Server Storage Advanced
Technology<br>
> > https://ecardfile.com/id/RobElliott<br>
> ><br>
> ><br>
> > *<br>
> > * For T10 Reflector information, send a message with<br>
> > * 'info t10' (no quotes) in the message body to majordomo at t10.org<br>
> ><br>
> <br>
> <br>
> <br>
> <br>
</tt></font>
<br>
--=_alternative 005CFDDD86256F6D_=--
More information about the T10
mailing list