RE: iSCSI: 2.2.6. Naming & mapping

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

Charles,
Though I think you are being very kind in your interpretation of Pierre's
note,  you may have a point.  Though I do not know why it is optional when
you have multiple connections -- the Spec does, however, say that the VPD
is optionally returned by the Inquire Command.  I think it is universally
applied when a Storage Controller has multiple connections, however, your
point may be -- that should we just state it as a non optional feature of
storage controllers that support iSCSI.  I can not see why this would be a
problem.  Does anyone else?



.
.
.
John L. Hufferd


Charles Monia <cmonia@NishanSystems.com> on 09/16/2000 03:31:57 PM

To:   Pierre Labat <pierre_labat@hp.com>, John Hufferd/San Jose/IBM@IBMUS,
      ips@ece.cmu.edu
cc:
Subject:  RE: iSCSI: 2.2.6. Naming & mapping



Apologies for the clutter if someone else has already responded to this.

> -----Original Message-----
> From: Pierre Labat [mailto:pierre_labat@hp.com]
> Sent: Thursday, September 14, 2000 5:57 PM
> To: John Hufferd/San Jose/IBM; ips@ece.cmu.edu
> Subject: Re: iSCSI: 2.2.6. Naming & mapping
>
>
> John Hufferd/San Jose/IBM wrote:
>
> > Folks,
> > We are again mixing issues and layers again.
> >
> > There is NO SUCH THING as an iSCSI LU.  There is an iSCSI
> device which is a
> > Controller which will have a SCSI layer which in turn
> supports an LU.  Once
> > the iSCSI session is established, the Rules of SCSI define
> how the LUs are
> > addressed.
>
> I agree.
>
> > There maybe additional Database (LDAP) processes and
> > information that attempts to relate LU #3 known to Host xyz
> to some name
> > "abcd...." etc.
>
> I don't want so much, i would want just to have a unique
> identifier per LU.
> It doesn't cost a lot and it will have the advantages
> described in the first
> mail below.
>

Hi:

As I read Pierre's note, he is proposing that support for the heretofore
optional Device Identification page be added to the list of features that
must be supported by an iSCSI device.  I believe that's the issue the ips
wg
ought to be debating.

(See ftp://ftp.t10.org/t10/drafts/spc/spc-r11a.pdf, section 8.4.3, for a
description).

Charles




> > But that is NOT an iSCSI Transport Protocol.  We may need
> > to work on this at some point but it is not an iSCSI
> transport issue.
> >
>
> Yes, it is not a transport issue, but why not request that
> now before people
> start
> building iSCSI controllers if it simplifies the life of everybody?
> I don't know how FC managed to have their WWN but why could not we
> do the same thing?
>
> Regards,
>
> Pierre
>
> >
> > .
> > .
> > .
> > John L. Hufferd
> >
> > Pierre Labat <pierre_labat@hp.com>@ece.cmu.edu on
> 09/14/2000 10:06:54 AM
> >
> > Sent by:  owner-ips@ece.cmu.edu
> >
> > To:   ips@ece.cmu.edu
> > cc:
> > Subject:  Re: iSCSI: 2.2.6. Naming & mapping
> >
> > julian_satran@il.ibm.com wrote:
> >
> > > Not again (what is the sign for frustration?)... I mean
> not before the
> > next
> > > version.
> > >
> > > Julo
> > >
> > > Raghavendra Rao <Jp.Raghavendra@EBay.Sun.COM> on
> 14/09/2000 21:04:02
> > >
> > > Please respond to Raghavendra Rao <Jp.Raghavendra@EBay.Sun.COM>
> > >
> > > To:   Julian Satran/Haifa/IBM@IBMIL
> > > cc:
> > > Subject:  iSCSI: 2.2.6. Naming & mapping
> > >
> > > I feel that a LUN should be very much part of the Naming scheme
> > > that has been proposed in the draft, regardless of the level of
> > > enforcement of the scheme by implementors. LUNs are very much
> > > part of SCSI addressing, and where LUN doesn't exist, zero is
> > > assumed.
> > >
> > > Without a LUN identifier, Naming section doesn't look complete.
> > >
> > > Do you have insights why it is omitted ? Is this by
> design or overlook ?
> > >
> > > Thanks.
> > > -JP
> >
> > About the topic of naming a LU, what seems to be  a benefit
> for me, is to
> > adopt the same requirement as fibre channel: each LU MUST provide a
> > unique identifier (from the Device Identification Page).
> > >From what i read it seems that it is not a big deal to add
> this page in a
> > LU.
> > However, this unique identifier is not needed for iSCSI
> protocol to work.
> > It doesn't interact with the protocol.
> >
> > But from an administration point of view, to configure a
> server using
> > storage
> > through iSCSI, the existence of this unique identifier helps a lot.
> > It allows the configuration product to know/check  if various LUNs
> > correspond
> > to the same LU, it could help to manage  the  LU migration
> (the LUN is
> > changed
> >
> > inside a target for the same LU).
> > It simplifies the configuration software by avoiding it to
> fake a unique LU
> > identifier.
> > It will help in having a configuration tool for iSCSI that
> can be closer to
> > the one
> > used with FC.
> >
> > Is somebody knowing if it is planned to incorporate this Device
> > Identification
> > Page
> > in the iSCSI LU? Will it be mandatory? Which format
> (identifier type) will
> > be
> > used?
> > FC  uses the type 3 (FC_PH Name_Identifier).
> >
> > Regards,
> >
> > Pierre
>