iSCSI-New draft

[julian_satran@il.ibm.com]

Dear colleagues,

I've submitted a new draft (I assume it will be named as requested
draft-ietf-ips-iSCSI-00.tx).


Things included:

   symmetric session with command status and data numbering
   support for shorter tags (for FC gateways)
   timers (for FC gateways)
   stateless mapping within gateways from iSCSI to FC and FC to iSCSI
   explicit logout
   data integrity & authentication support
   per LU target tags
   support for immediate and non-immediate unsolicited data
   a mechanism for soft framing using the urgent pointer
   2 types of resets (although a bit different from what we agreed - it did
   not work out well with the C-bit)
   support for Bi-Directional commands (although as kludgy as in the FCP
   proposal but I am still working to get it in a more decent shape)


Things not included:

   third party logout (no compelling arguments for not relegating it to a
   managing entity that should do fencing)
   a deliberate separation into an FCP and iSCSI header - beyond other
   considerations it would have tied us up to FCP forever - not a wise
   thing to do. However mapping is a no-brainer.


The draft is heavily revised and we hope that it is better in answering
both the "how" and the "why" questions although it is obviously not meant
for the uninitiated.

I will continue doing some more editing during the next two weeks to add
more examples, state diagrams (if required) and things we have overlooked.

Thanks the whole list for their active contributions. I hope I acknowledged
all the active participants - and if not please accept my apologies in
advance.

And thanks Costa Sapuntzakis ,  Matt Wakeley and Kalman Meth - who beyond
their technical contributions have also made the current draft to look more
like an American-English text than to an Hebrew-English one.

Regards,
Julo

Julian Satran - IBM Research at Haifa