iSCSI: draft review - data pdf numbering

[Matt Wakeley <matt_wakeley@agilent.com>]

Section 1.2.2 and 1.2.2.3, Data PDU numbering.

I don't see the need for the "Data numbering".  In the extremely
unlikely event a TCP connection fails, the command should be
restarted in it's entirety.  Otherwise, the initiator will have
to piece together data from a failed I/O with incomplete data from
a retried I/O and hope it got the data right.  Too much complexity
for an event that shouldn't happen very often.

Section 1.2.2.3 Data PDU numbering

(if Data PDU numbering is supported) NOP commands must carry a
Target Task Tag for the target to match up the "DataRN" with the correct I/O.