stds-1394: RE: SBP-2 Study Group

Peter Johansson PJohansson at acm.org
Fri Aug 11 16:24:09 PDT 2000


* From the T10 Reflector (t10 at t10.org), posted by:
* Peter Johansson <PJohansson at ACM.org>
*
At 03:56 PM 8/11/00, Robert Snively wrote:

>I am a little bit confused about what might constitute peer to peer 
>behavior.  If I understand generic computer architecture, one program 
>wants to do something and asks another program to do it on its behalf.

Yes, a client and a server.

>SCSI is peer to peer in that any end of a link can choose to be an 
>initiator and ask any end to be a target.  However, the invocation of a 
>function is an initiator operation, and the execution and management of 
>the requested function is a target operation.  If SCSI soon allows 
>bi-directional data transfer, it is still target managed.

There are some important distinctions you're eliding here. The endpoints 
are typically manufactured with either initiator or target (or both) 
capabilities. Only the device manufactured with both initiator and target 
capabilities can "choose".

But why should invocation of a function necessarily be an initiator 
function? What if a client resides at a target and a service resides at an 
initiator? This is the work that has gone into the Peer to Peer Data 
Transport (PPDT) protocol, aka IEEE P1394.3. It makes it easy for either 
initiator or target to request the formation of a bi-directional 
connection. Call it a pipe or one or more sockets, if you prefer. The 
application client and server are able to remain unaware as to their 
execution environment, SBP-2 initiator or target, as the PPDT abstraction 
hides those details.

>So what do people mean when they say peer to peer as if it were somehow 
>different from SCSI?

What I described above is one of the things people mean when they say "peer 
to peer" and it seems different, at least to me, from SCSI.

There is a lot of muddy thinking about peer to peer, so maybe some people 
have something else in mind as well.




Regards,

Peter Johansson

Congruent Software, Inc.
98 Colorado Avenue
Berkeley, CA  94707

(510) 527-3926
(510) 527-3856 FAX

PJohansson at ACM.org

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




More information about the T10 mailing list