Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!uwm.edu!bionet!agate!ucbvax!minestrone.Berkeley.EDU!gnn
From: gnn@minestrone.Berkeley.EDU (George Neville-Neil)
Newsgroups: comp.arch
Subject: Re: Networking for Distributed Computing
Message-ID: <41616@ucbvax.BERKELEY.EDU>
Date: 6 Apr 91 00:30:55 GMT
References: <1991Apr5.182853.20728@hubcap.clemson.edu>
Sender: nobody@ucbvax.BERKELEY.EDU
Reply-To: gnn@minestrone.Berkeley.EDU (George Neville-Neil)
Organization: Lawrence Livermore National Laboratory
Lines: 47

In article <1991Apr5.182853.20728@hubcap.clemson.edu>,
mccalpin@perelandra.cms.udel.edu (John D. McCalpin) writes:

|>The system I envision would consist of some moderate number of
|>high-performance cheap workstations (IBM RS/6000-320 or HP/9000-720),
|>lets take 32 as an example.  An additional SCSI interface on each unit
|>would provide 7 high-speed, point-to-point interfaces, which could be
|>reconfigured (via a patch panel) to provide a large number of
|>different topologies, including: a line, a 2-D mesh, a 3-D mesh, a
|>ring, etc.

|>So what is wrong with this idea?  Am I misunderstanding what can be
|>done with a SCSI interface or how hard the implementation of a
|>buffered FIFO would be?


Hmm.  I am not sure about this idea.  I am not an expert on SCSI but I seem to
remember that you can only have seven (maybe eight) SCSI devices on a
SCSI string
(one cable/controller).  Also you would be limited by transmission distance.  

If it is true that the processors/systems/whatever would have to be within a
close proximity (a meter or two) wouldn't it be cheaper to build something like
this using cards in a backplane ??  

Also, the SCSI spec was basically built around disk architectures and
synchronizing reads and writes etc.  If this is all you want the net to do then
fine.  But this also means that if you want TCP/IP or UDP or some other
protocol
it will have to be done in software with the SCSI stuff as a low level
protocol in your network.  It's doable but is it worth it ??

To sum up, I'm not sure, but I don't think it's an optimal approach.

You might look into the literature to see if anyone has used SCSI as a
network instead of as just a master/slave controller system.

Hope this helped....

Later,
George


George Neville-Neil      		Kinky is as kinky does.
gnn@mammoth.berkeley.edu 

Warning: Taking drugs approved by the FDA can be hazardous to your health.