Re: Framing (was RE: iSCSI: new draft)

[Matt Wakeley <matt_wakeley@agilent.com>]

Marjorie,

You are starting to dive into implementation.  From outside the box, TCP has to
"look" like it's defined in the standard.  How that is achieved is up to the
implementor.

My whole point of this discussion is that the standard is supposed to define
how the protocol works from outside the box.  How SCSI/iSCSI/TCP communicate
and operate with each other inside the box must not be specified.

-Matt

"KRUEGER,MARJORIE (HP-Roseville,ex1)" wrote:

> ..snip..
> >>  And if it did, what will it do in the TCP layer to
> >> indicate those "out of order" bytes have already been delivered?
> >
> > It won't be able to do anything in the TCP layer to indicate
> > those bytes have been delivered.
>
> Is this where you are thinking of using some sort of "score boarding"
> to keep track of what bytes in a receive buffer have been "resolved" and
> what bytes are currently outstanding?  Cause if TCP doesn't
> somehow know these "out of order bytes" have
> been moved "to the correct place",  won't it "deliver" them
> again to iSCSI when the missing segment is recovered?
>
> > That would break the existing TCP model.
> > The only thing it could do is use the SACK option to help the sending TCP
> > reduce the amount of retransmitted data.
>
> Does "it" refer to TCP or the iSCSI intervention?  I assume you mean "an
> iSCSI/TCP optimized adapter would use SACK"...?
>
> > -Matt Wakeley
> > Agilent Technologies
> >
>
> -Marjorie