|
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] Re: iSCSI: definitionsEddy, Apologies, this reply is a bit out of order (I've only been doing a scattered job of handling e-mail this season). I think there's a "perspective" difference going on here. First, SAM has defined "abstract" structures, not real things. It is the protocols that specify the "implemention" of the abstract model defined by SAM. In light of this, you're third question is phrased backwards! iSCSI defines the iSCSI Target Node name, then maps that to the SCSI target device name. As in my other note, iSCSI defines its concrete architecture and then maps those to SCSI stuff. Second, (as you note in your first point) there are two different generic definitions for the term "initiator" and "target" and it depends on which document you're reading. If you read SAM (or any T10 document), the term "initiator" is a synonym for "SCSI initiator port" and "target"="SCSI target port". See the glossary of SAM-2. However, in the iSCSI draft (and I think it says this somewhere in the glossary or early in the document), the term "initiator" is a synonym for "iSCSI Initiator Node" and "target"="iSCSI target node". *I believe this is one source of the confusion we're seeing!*. To be consistent with SAM-2, the iSCSI document would have a major "Find and Replace" job. I have no objections to that. (In all my notes and text in the iSCSI draft, I have tried to never be ambiguous but to fully qualify the terms in every case -- though I may have failed in some cases :-{)) We don't specifically define either the term "iSCSI Initiator" or the term "iSCSI Initiator Portal Group". However, we do define the term "iSCSI Node" and "iSCSI Portal Group" -- and these can both come in "initiator" flavor as well as "target" flavor. Perhaps we need to be more pendantic. Jim Hafner Sent by: owner-ips@ece.cmu.edu To: Julian Satran/Haifa/IBM@IBMIL
Home Last updated: Wed Jan 02 01:17:49 2002 8244 messages in chronological order |