[comp.periphs.scsi] SCSI A/B/C switcher

dyer@spdcc.COM (Steve Dyer) (09/02/90)

Has anyone ever come across a multi-way SCSI port switcher?  For something
like sharing an Exabyte between multiple heterogeneous systems, it would
seem to be perfect.  For example, I've got an Exabyte which I'd like
to use on an RT, a Mac and a Sun.  All 3 machines are in close proximity
to each other.  Rather than having to fiddle with the cabling each
time I was to use it on a different machine, I could simply power down
and flip a switch.  Much less stress on the cables and connectors.

Is there any reason why such a device hasn't been built?  It needn't be
a cheap-o box--I'm willing to pay for the proper quality connections
and switches...


-- 
Steve Dyer
dyer@ursa-major.spdcc.com aka {ima,harvard,rayssd,linus,m2c}!spdcc!dyer
dyer@arktouros.mit.edu, dyer@hstbme.mit.edu

chugins@hpcupt1.HP.COM (Chris Hugins) (09/04/90)

I believe that SCSI will have severe problems with a switch-box.  There are
several reasons for this, not the least of which is the "glitching" which
would occur on the bus.  Many devices' transceivers may "glitch" the bus
at power-on and/or power-off.  No CRC is employed on data, therefore any
other device active on the bus may have the data being transfered be
corrupted in an undetectable manner.  Some manufacturers have recently
been using "glitchless" transceivers so to address this problem.

SCSI expects its bus to be terminated at each end.  "Breaking" a SCSI
bus with a switch-box may give interesting results.

You may dedicate a SCSI bus to your shared device.  Thus, you would have
no problems with other devices being active on the bus.  Also, you would
know that I/Os are quiescent to this device when you are messing with the
switch.  The processors that are not switched to the device would have
unterminated SCSI busses, but they wouldn't be used at that time.

That does waste some connectivity however.

Chris T. Hugins
chugins@hpisoa2.hp.com