Date: Wed, 16 Jul 2003 13:47:55 -0500
From: Caitlin Bestler <cait@asomi.com>
Subject: Re: iSCSI: Login Redirect
To: John Hufferd <hufferd@us.ibm.com>
Cc: ips@ece.cmu.edu, "Mallikarjun C." <cbm@rose.hp.com>,
	Lakshmi Ramasubramanian <nramas@windows.microsoft.com>,
	owner-ips@ece.cmu.edu
In-Reply-To: <OF6D1BD911.CFF612E9-ON88256D65.0046C0CD-88256D65.0047DC4A@us.ibm.com>
Message-ID: <r02000000-1026-0375BA08B7BE11D7838C003065D48EE0@[192.168.0.2]>
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: owner-ips@ece.cmu.edu
Precedence: bulk

On Wednesday, July 16, 2003, hufferd@us.ibm.com wrote:

>Extending my own note... I also believe it would be wrong to
>set a minimum number of guaranteed redirects.  This is because,
>I expect several system vendors to use iSCSI as a DAS
>technology as well as SAN technology. (Similar to the way a
>number of vendors "Racked" their FC external Raid controllers
>together with the server as a Direct Connection.)  When used in
>this manner the customer does not even need to think about the
>system being on a SAN.  Yet the Storage/System Vendor could use
>the very same Raid controller, which they use in a SAN, as a
>direct attachment to the Server in a RACK.  In this environment
>if any redirect occur, it might be considered a error.
>
>So in the above DAS environment having a minimum is, IMHO, not 
>approprate.
>
Very good point. Any minimum needs to be expressed relative to
a deployment profile, and definitely be no stronger than a SHOULD.
Some guidance on what a *typical* initiator deployed in a general
environment should support would still be a good idea to state.