RE: Towards Consensus on TCP Connections

[Peter Johansson <PJohansson@acm.org>]

At 10:24 AM 8/14/00, Douglas Otis wrote:

>Firewire has many strikes against it for broad acceptance.

Not the least of which is the surprising persistence of outdated and 
inaccurate information.

>It uses animbalanced pair to identify the PHY.

I'm not sure I understand your assertion---nor what you perceive as a problem.

>Licensed PHY from Apple at a cost per port with mode use restricted.

Not true. Various owners of intellectual property pertinent to IEEE 1394 
formed what is informally known as the "patent pool"; the 1394 Licensing 
Authority makes licenses available on reasonable terms. Not only are the 
charges NOT per port, as you say, they are per INTEGRATED SYSTEM (not even 
component) and at a flat rate per system.

>Limited range of the copper solution does not satisfy most commercial 
>environments.

Partially correct, but applicable only to the DS-encoding over copper 
specified by IEEE 1394-1995. The diversity of media specified by P1394b 
(glass fibre, plastic fibre, UTP-5) provide point-to-point lengths in the 
hundreds of meters. Since this is the same geographical range proposed by 
interconnects such as Infiniband or gigabit Ethernet, it seems suited to 
some commercial environments.

>No aggregation as Firewire is a half-duplex daisy chain.

Please explain more; I do not understand your point. What is being 
aggregated? Or failing to be aggregated? Give me a concrete example.

>Although possible to use as a typical SCSI target with IP, it is not for 
>storage.

IEEE 1394 is well-suited to storage, as evidenced by the performance 
characteristics of SBP-2 disks. As previously mentioned, though, neither 
SBP-2 nor other "native" SCSI transports (such as FCP) are compatible with 
the group's direction with iSCSI. But that has nothing to do with the 
underlying utility of IEEE 1394 of Fibre Channel as an IP transport which 
is, in turn, used by iSCSI.

>... implementing any IP standard on any medium would be encouraged.

Agreed. See RFC 2734 for the details.




Regards,

Peter Johansson

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

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

PJohansson@ACM.org