Project Proposal for SCSI-4 Parallel Interface - 3 (SPI-3) Physical & Protocol Layers

[Gene Milligan]

* From the T10 (formerly SCSI) Reflector (t10 at symbios.com), posted by:
* Gene Milligan <Gene_Milligan at notes.seagate.com>
*
<<Project Proposal for SCSI-4 Parallel Interface - 3 (SPI-3)>>

 I recommend that this be Project Proposal for SCSI Parallel Interface - 3 
(SPI-3). For some of us SCSI embraces or at least allows a variety of 
transports all supporting the SCSI command sets and architecture (1394 at least 
can be construed to do this and may very well in the fullness of time actually 
do it).

 While I enjoy original satirical banter, I think the project itself should 
avoid the cutesie <<with the adoption of a 'low-fat' protocol (LFP) >> in terms 
of nomenclature.

 I think << and provide unrestricted hot-plugging for parallel SCSI>> is a good 
goal which should be pursued but left out of the project proposal. Inclusion in 
the standard should be upon completion of the design tradeoffs and testing.

 <<SPI-3 is will also deal with the advancing developments in silicon 
technology related to power management and voltage-reduction.  These are key 
features of the next generation of
chips.>> Aside from the typo this is also something worthwhile to at least 
consider but "dealing with" is what politicians do. The statement should either 
be more definitive or preferably left out as one of the items falling under 
item (c) of the scope.

 My comments on the scope will be minimized by changing <<The SPI-3 standard 
will:>> to "The SPI-3 project will:".

 <<a) increase the data rate to 80 mega-transfers per second.>> I to presently 
favor this rate but think the project proposal at this point should be less 
specific. "at least 80" would be one way to be less specific.

 <<The SPI-3 standard is intended for use in open systems.>> This seems to be 
the opposite of what we usually state and invites a definitive answer on the 
relationship is to the open systems model. Why is this not  intended still for 
closed systems?

 << Invite comments and proposals from organizations that may have a 
contribution to the SPI-3 standard.>> This may be the wording T10 has used in 
the past but I think it actually should be " Invite participation, comments, 
and proposals from organizations that may be materially impacted by the SPI-3 
standard."

 Sorry I have asked before but forgotten (no offense intended) which 
<<government agencies>> are currently members of T10?

 I have also commented in the past that I do not think we should bind ourselves 
to either the minimum or maximum working group meeting called out in <<for one 
to three days>>. If the Chair disagrees then we should have an emergency 
reschedule of the November meetings which are less than a full day.

 Rather than << A Call for Patents will be made.>> I think the practice is 
"Calls for Patents will be made."

 At first I wondered why <<             SCSI Primary Commands - 2        1236-D 
and 
             SCSI Architecture Model - 2      1157-D>> are listed under related 
development projects. Then I decided there was a reason. So their predecessors 
should be listed under existing standards. In both cases based upon the work 
already done FCP should also be in both places even though those not familiar 
with the detailed proposals may be a bit puzzled.

 <<It is anticipated that SPI-3 will be fast tracked for publication as an ISO 
standard.>> It is true that Fast Track is an option with tradeoffs of skipping 
one step in the process but starting the process significantly later. So it is 
not Fast Arrival. There is no need for the domestic project to be specific on 
which of the ISO processes will be followed and sufficient to just say "be 
submitted".

 <<None>> is not correct for other related ISO projects unless it had been 
correct for domestic projects.

 << None.>> is probably also incorrect for close liaison if FCP becomes an 
option of SPI-3. I also wonder but certainly don't insist whether or not we 
should have liaison with other standards groups developing standards for the 
parallel interface on graphics products.

Gene



*
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at symbios.com