RE: iSCSI Naming and Discovery

["John Hufferd/San Jose/IBM" <hufferd@us.ibm.com>]

Joshua,
I think the LU being part of the "/Modifier" is a bit problematical.  You
should focus on the connections each end, and let the LU stuff be
transparent to your Gateway.  They will of course use Report LUNs and
Inquiry and use their approprate LU number, I am having problems
understanding why a gateway would need to know the LU number.

Could you resend your note without the LUN stuff so that we can see, a
little easier, what you think is really needed.

.
.
.
John L. Hufferd
Senior Technical Staff Member (STSM)
IBM/SSG San Jose Ca
(408) 256-0403, Tie: 276-0403
Internet address: hufferd@us.ibm.com


Joshua Tseng/Nishan Systems <joshua.tseng@NishanSystems.com> on 10/05/2000
08:43:46 AM

To:   John Hufferd/San Jose/IBM@IBMUS, ips@ece.cmu.edu
cc:
Subject:  RE: iSCSI Naming and Discovery



Hi John,

>Joshua Tseng,
>I think it is the job of the Gateway/Bridge/Router to have a map between
>the iSCSI initiator identification to a logical WWN, there are a lot of
>ways to do this,  but regardless, it is a requirement to make that WWN
>mapping known to Administrative Software and Storage Administrators.  The
>Storage Administrator will then need to tell the Storage Controller what
>LUs to permit access from that WWN.  Jim Hafner and Costa have been
>suggesting that a Double address is needed to connect.  Though they are
>suggesting it for IP based Gateways, (which you may or may not agree with)
>would it be approprate and how would you use it in your iSCSI to FC
>Gateway?

I hope I understand your question correctly...

In my opinion iSCSI needs a way to represent basic SCSI and Fibre
Channel naming constructs.  I support the URL naming proposal using the
which includes the WWN and LU in the <modifier> portion of the URL.

Yes, as you state the Gateway/Bridge/Router has a map between between
the iSCSI name and the FC WWN, but what is that "iSCSI initiator
identification"?  Do we need to add another layer of virtualization
to translate the WWN into the iSCSI equivalent?  Or can't we just
use the WWN and LU identifiers as they exist in SCSI and FC?  I think
we need to make the Gateway/Bridge/Router as stateless as possible,
which means preserving as much of the FC naming conventions as possible
for use in the iSCSI world, reducing the amount of virtualization
that the Gateway/Proxy has to do in translating between FC and iSCSI
worlds.

Same thing with LU's.  When an iSCSI initiator sends an INQUIRY command
to LUN 0 of a FC target, the Gateway/Proxy shouldn't have to intercept
the response to translate the LU masking information to the "iSCSI
equivalent".  When an iSCSI initiator receives the LUN masking
information, that initiator should be able to directly use
the LU information and address those LU's.

Josh