SORT BY:

LIST ORDER
THREAD
AUTHOR
SUBJECT


SEARCH

IPS HOME


    [Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

    Re: iSCSI: naming/LUN views



    Mark,Jim,
    
    
    Mark Bakke wrote:
    
    > Pierre-
    >
    > Sorry I took so long to reply; for some reason, I didn't receive
    > your original message from the reflector.
    >
    > I agree with all of Jim's comments, but would state this one
    > more strongly:
    >
    >   A TargetWWUI MUST present exactly the same LU view to an
    >   InitiatorWWUI regardless of the portal on which it is accessed.
    >
    > This means that we have a method of identification at the target
    > level, which we were missing (at least as a standard) from FC and
    > parallel SCSI.
    
    Perfect.
    
    >
    >
    > This was our intent, and would solve the problem you stated of
    > having to compare views.  Perhaps we should add this statement
    > to the document.
    
    Yes, i think it has to be said explicitely.
    
    Regards,
    
    Pierre
    
    >
    >
    > --
    > Mark
    >
    > Jim Hafner wrote:
    > >
    > > Pierre,
    > >
    > > In my view, one should get the same LU view in your situation because there
    > > is effectively no other information by which the target might present a
    > > different view.  Namely, the target (that creates/enforces the view) is the
    > > entity scoped by the TargetWWUI.  It uses the Initiator WWUI to construct
    > > that LU view.  That might be constructed by rules established via SCSI
    > > access controls (and yes, the InitiatorWWUI is the right mapping to
    > > TransportID though that has not been formalized anywhere yet).
    > >
    > > This has some interesting implications.  Within the context of a single
    > > SESSION established through multiple connections, there is only the one I_T
    > > nexus.  But we can also establish multiple sessions between the same
    > > InitiatorWWUI and TargetWWUI.  As far as access controls are concerned, the
    > > views should be identical.  As for other properties of I_T nexus', e.g.,
    > > reservations, the jury is still out.
    > >
    > > Jim Hafner
    > >
    > > Pierre Labat <pierre_labat@hp.com>@ece.cmu.edu on 02-27-2001 10:38:26 AM
    > >
    > > Sent by:  owner-ips@ece.cmu.edu
    > >
    > > To:   ips@ece.cmu.edu
    > > cc:
    > > Subject:  iSCSI: naming/LUN views
    > >
    > > Mark (Bakke),
    > >
    > > After going through the naming/discovery drafts,
    > > a question came to my mind:
    > >
    > > Does several connections flowing through
    > > several portals but using  the same initiator WWUI
    > > and target WWUI will see the same thing on the target?
    > > I mean by same thing: same LUs associated to the same LUNs.
    > >
    > > I guess that the initiator WWUI acts as an transportID (SCSI access
    > > control)
    > > and gives the same LUs view independently of the portal used.
    > > Is it the case?
    > >
    > > If it is not the case, the entity that creates multiconnections
    > > sessions
    > > will need out of band information to know if the  2 portals can be used
    > > for a same session or will have to do the  painfull job to compare the
    > > LU views
    > > from the two portals with no guarantee that the views could not change
    > > in the future.
    > >
    > > Regards,
    > >
    > > Pierre
    >
    > --
    > Mark A. Bakke
    > Cisco Systems
    > mbakke@cisco.com
    > 763.398.1054
    
    


Home

Last updated: Tue Sep 04 01:05:30 2001
6315 messages in chronological order