Proposal for IEEE company_id based formats for FC-PH world-wideidentifiers
Mike Wenzel
mw at core.rose.hp.com
Fri Jan 10 16:18:39 PST 1997
* From the SCSI Reflector (scsi at symbios.com), posted by:
* Mike Wenzel <mw at core.rose.hp.com>
*
Hi Rod and All,
Once again, I'm pretty happy with the outcome of the SCSI meetings on this.
I really like the examples and would like to post the .pdf on the Serial
Solutions web page. Thanks for working these up!
> Mike and I both agree that the company_id and vendor specified id of
> the FRU controller should NOT be used in constructing the LUN WWNs.
>
> I would go even further and require that all LUN WWNs for a RAID
> controller complex have the same company_id and vendor specified id
> fields (even across FRU controller swaps).
I sort of agree with both of you:
A. All of us agree that the Volume WWN must be permanently attached to
the chunk of storage as long as it retains its identity.
- As an aside, I REALLY think the Volume WWN needs to change when a
Volume loses its previous identity by reformatting, etc. I want
hosts to be able to detect when a LUN ID results in access to a
logically different chunk of storage than what the host might have
expected (e.g., if it didn't know about the reformatting). This
will really be important for arrays that can be dynamically
reconfigured in terms of their LUN ID and Volume relationships
wrt underlying storage.
B. I agree with Larry in that if the WWN is "attached" to a controller
FRU, then it could cause big trouble in ensuring that Vendor Extension
values are not re-used, when the controller is moved around. If a
vendor thinks he can handle this to where WWNs are always unique,
regardless of how the controller FRU is moved around, then I agree with
Rod's examples completely. If, as I suspect, this is too hard to
implement, then it would be better if something else were used for the
seed to create the Volume WWN. Maybe part of the enclosure is likely
to be more permanent. Maybe the array has its own WWN just as a
Volume name seed.
- By the way, MUST controllers have WWNs attached? I know a FibreChannel
"Node" has a WWN that's used at login. Has this been equated with
a controller FRU in the SCSI discussions or could a vendor attach
the Node WWN to the FRU "slot" and identify the FRU by some SES
mechanism? In the latter case, the Node WWN would not change when
the controller is replaced.
C. I think it's unnecessarily constraining to REQUIRE all the Volume WWNs
in an array to have the same value in the first 64 bits of the 128b
format. I expect that they normally WILL be the same in most cases,
because that's probably the easiest way to manage the WWNs.
But I don't want to require it. For example, I wouldn't want a host
to depend on this name structure to find all the Volumes in a given
array enclosure! This could really pinch somebody at some point.
Was there any more discussion about the list one the Device ID page and
under what cases there should be more than one WWN on the page? This
would be an excellent addition to the examples you've already built.
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