Re: Bidirectional SCSI commands and iSCSI

[Matt Wakeley <matt_wakeley@agilent.com>]

or 3) make the header *NOT* contain the CDB.  Instead, let the CDB be part of
the iSCSI command message payload (like it is in FC).

-Matt

Jim Hafner/Almaden/IBM wrote:

> Folks,
>
> I'm not advocating variable length headers here to support bi-d commands.
> More to the point, I'm suggesting that the fixed length header change (NOW)
> to accomodate the bi-directional SCSI commands. Better now while the draft
> is still in flux than later after its finalized and more hardware gets
> built.
>
> BTW, I see two approaches:
> 1) one universal header with enough fields for both directions
> 2) one header for classical uni-d commands and an additional header for
> bi-d commands (as FCP is proposing)
>
> My vote would be for option (1).
>
> Jim Hafner
>
> Stephen Bailey <steph@cs.uchicago.edu>@ece.cmu.edu on 10-03-2000 07:12:46
> AM
>
> Sent by:  owner-ips@ece.cmu.edu
>
> To:   ips@ece.cmu.edu
> cc:
> Subject:  Re: Bidirectional SCSI commands and iSCSI
>
> > I think that software initiators will be widely used and they are far
> > more efficient with fixed length headers for widely used operations.
>
> In ST, fixed length headers were a design choice to permit hardware
> acceleration.
>
> Other than on very slow links, it's hard to argue with the performance
> advantages of fixed length headers under any circumstances.
>
> Steph