Proposal for IEEE company_id based formats for FC-PH world-wideidentifiers

DeKoning, Rod rdekonin at ppdpost.ks.symbios.com
Fri Jan 10 11:59:00 PST 1997


* From the SCSI Reflector (scsi at symbios.com), posted by:
* "DeKoning, Rod" <rdekonin at ppdpost.ks.symbios.com>
*



Mike, Larry, et.al.,

I just got back from the SCSI meetings yesterday afternoon.  We touched on 
this topic some during Bob Snively's discussion of the WWN tutorial.  During 
that discussion, it was clear that several people in the group did not fully 
understand the use of the IEEE 128 bit WWN usage, so I will convey this 
information to you and the reflector with hopes to dispel the confusion. ( 
The following description was verbalized at the meeting and helped those who 
did not previously understand the topic.)

The key to this discussion is that the IEEE WWN assigned to the Logical Unit 
must be treated as a permanent identifier of the logical unit (more aptly 
named Volume, since Logical Unit Numbers become nothing more than another 
component of the path to the storage volume.).

Furthermore, the 128 bit WWN must be considered as one 128 bit identifier 
rather than the two individual identifiers concatenated together.  Another 
way to look at this is to view the controllers WWN (at creation time) as the 
seed for the creation of the volume's (LUN's) WWN.  Once the volume has a 
WWN, the controller's WWN and its association with the volumes WWN becomes 
irrelevant.

I hope the following examples further illustrate the permanence of the WWN 
assigned to a volume (LUN)

(Note: I hope to get an illustrated version of these examples in .pdf format 
on a publicly accessible ftp site early next week.  Until then, the key is 
to look at the least significant byte of the WWNs in the following examples. 
 I have tried to keep the other bytes consistent to aid in folloiwing the 
example.)

 ----------------------------------------------------------------------------  
 ----------------------------------------------------------
 ----------------------------------------------------------------------------  
 ---------------------------------------------------------



Example #1: Using one controller which is a FRU (borrowed from Bob's 
tutorial).

Originally we have the following configuration:

Controller Company ID = AC DE 48
Controller's Vendor Specified Identifier for THIS Controller =
               B 17 34 F6 2D

Therefore, the WWN for this Controller (using the 64 bit format) would be:
                5A CD E4 8B 17 34 F6 2D


_______________

Now lets say someone creates a volume (LUN) on this controller:

The controller chooses the following for THIS volumes Vendor Specified 
Region of the WWN:
                  12 34 56 78 9A BC DE 31

Therefore, the WWN for this volume (LUN) connected to this controller is:
                  6A CD E4 8B 17 34 F6 2D    12 34 56 78 9A BC DE 31

______________

Now lets say the controller as described above fails and a new controller is 
swapped in its place.

The new Controller Company ID remains the same in this example = AC DE 48
Controller's Vendor Specified Identifier for THIS Controller (note that I 
only change the least significant byte for the example)
                                = B 17 34 F6 95

Therefore, the WWN for this Controller (using the 64 bit format) would be:
                            5A CD E4 8B 17 34 F6 95

The volume that previously was defined remains with the EXACT SAME WWN. 
 i.e. the WWN for this volume (LUN) connected to this controller is:
                  6A CD E4 8B 17 34 F6 2D    12 34 56 78 9A BC DE 31

_______________

Now lets say someone creates a 2nd volume (LUN) on this controller:

The controller chooses the following for THIS volumes Vendor Specified 
Region of the WWN:
                  12 34 56 78 9A BC DE 98

Therefore, the WWN for this 2nd volume (LUN) connected to this controller 
is:
                  6A CD E4 8B 17 34 F6 95    12 34 56 78 9A BC DE 98


Note that the first LUN's WWN remains attached to this controller, and keeps 
its original WWN.

_______________________________________
_______________________________________

EXAMPLE #2 Dual Controller Management

Originally we have the following 2 controller configuration both attached to 
the same storage:


Controller #1 Company ID = AC DE 48
Controller's Vendor Specified Identifier for Controller #1 =
               B 17 34 F6 2D

Therefore, the WWN for Controller #1 (using the 64 bit format) would be:
                5A CD E4 8B 17 34 F6 2D


Controller #2 Company ID = AC DE 48
Controller's Vendor Specified Identifier for Controller #2 =
               B 17 34 F6 54

Therefore, the WWN for Controller #2 (using the 64 bit format) would be:
                5A CD E4 8B 17 34 F6 54


_______________

Now lets say someone creates a volume (LUN) using controller #1:

Controller #1 chooses the following for THIS volumes Vendor Specified Region 
of the WWN:
                  12 34 56 78 9A BC DE 63

Therefore, the WWN for this volume (LUN) connected to BOTH controller #1 and 
controller #2 is:
                  6A CD E4 8B 17 34 F6 2D    12 34 56 78 9A BC DE 63


_______________

Now lets say someone creates a 2nd volume (LUN) using controller #2:

Controller #2 chooses the following for THIS volumes Vendor Specified Region 
of the WWN:
                  12 34 56 78 9A BC DE 78

Therefore, the WWN for this volume (LUN) connected to BOTH controller #1 and 
controller #2 is:
                  6A CD E4 8B 17 34 F6 54    12 34 56 78 9A BC DE 78

Volume #1 is still accessible through both controllers at its WWN.


END of Example



As far as FRU information is concerned, this could potentially be solved 
using SES.  Either way, I think it is a separate issue, and does should not 
be linked to WWN discussions.  I completely agree with Mike's statement that 
we do not want to get into the mess of attempting a 2nd way of identifying 
FRUs using the Device Identification page.  Furthermore, as I think the 
above example illustrates, we do not want to tie FRUs to volumes (LUNs) for 
the reasons previously stated.


Regards,
Rod DeKoning
Symbios Logic





 ----------
>From: Mike Wenzel
>To: Larry Chen
>Cc: Mike Wenzel; scsi; serial_solutions
>Subject: RE: Proposal for IEEE company_id based formats for FC-PH 
world-wideidentifiers
>Date: Wednesday, January 08, 1997 5:41PM
>
>
>*
>* From the serial_solutions Reflector, posted by:
>* Mike Wenzel <mw at core.rose.hp.com>
>*
>* To post to this Reflector, send email addressed to
>serial_solutions at zk3.dec.com
>*
>
>Hi Larry,
>
>At 12:47 PM 1/8/97 PST, Larry Chen wrote:
>>Hi Mike,
>>
>>I thought one of should send out email to record our recent conversation 
in
>>this matter. I hope you don't mind if I use my notes.
>
>Good idea.  The conclusions we came to are probably of interest to others
>as well.
>
>>Everyone agrees that the LUN WWN should be invariant (which happens to 
make
>>the company_id + vendor specified id fields invariant too). So actually,
>>using the company_id + vendor specified id fields for the RAID controller
>>WWN is NOT ambiguous at all (and may not provide any additional utility).
>
>Yes, this is VERY important and I'm glad we all keep coming back to the
>same conclusion!
>
>In my previous memo, I said I was worried about using the WWN of a
>controller as the base for the WWN for a LUN, when the controller is
>a field-replaceable unit (FRU).  I should have clarified that this
>assumes that the WWN of the controller stays with the controller FRU
>as it is moved around.  If the WWN stays with the "slot" that the
>replacement plugs into, then it is invariant wrt the LUN and should
>pose no problems with ambiguity.
>
>>If the RAID controller's FRU WWN is needed, then there were two suggested
>solutions:
>>
>>1) Use the Device ID page - add FRU associations
>
>Rod DeKoning made a proposal some time back where each item in the list
>on the Device ID page would have an "association" field that could name
>the entity being identified by that item.  This field does not appear in
>either of Bob's tutorials or the most current SPC document.  I'm not sure
>what became of the proposal, but it would have to be re-discussed to
>pursue this alternative.
>
>>2) Use the ASCII Information page - on further observation, this page does 

>>not provide the WWN of the FRU (but does have a Vendor-specific 
information
>>field which could be used for this purpose).
>
>It would be great if we could use standard mechanisms to locate all the
>FRUs in a path to a chunk of storage.  Vendor-dependent mechanisms should
>be a last resort.  I'd like to see this discussed by the Serial Solutions
>group.  It's probably time to put together another agenda...
>
>>
>>Bye for now.
>>Larry
>
>
>Best Regards,
>
>Mike
>
> ************* |  Mike Wenzel,
> *****   ***** |  Hewlett-Packard - NCD System Interconnect Lab,
> *** /_  _ *** |  Mailstop 5601,
> ** / / /_/ ** |  8000 Foothills Blvd.,
> ***   /   *** |  Roseville, CA 95747-5601
> *****   ***** |  E-mail: mw at core.rose.hp.com
> ************* |  Telephone (916) 785-5609  FAX (916) 785-2875
>
*
* For SCSI Reflector information, send a message with
* 'info scsi' (no quotes) in the message body to majordomo at symbios.com




More information about the T10 mailing list