Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!swrinde!zaphod.mps.ohio-state.edu!sol.ctr.columbia.edu!ira.uka.de!smurf!urlichs
From: urlichs@smurf.sub.org (Matthias Urlichs)
Newsgroups: comp.periphs.scsi
Subject: Re: *More addresses for SCSI bus***
Message-ID: <07kug2.0f5@smurf.sub.org>
Date: 9 Dec 90 17:52:12 GMT
References: <1990Dec5.115702.7605@swift.cs.tcd.ie>
Organization: University of Karlsruhe, FRG
Lines: 36

In comp.periphs.scsi, article <1990Dec5.115702.7605@swift.cs.tcd.ie>,
  bhnnessy@swift.cs.tcd.ie writes:
< 
<I am investigating the viability of a RAID(Redundant Array of Inexpensive Disks)
<system. This would require the connection of about 20 disks. However only a few
<SCSI addresses are available to connect the system. To get around this problem
<I was wondering if it is possible to buy a SCSI device that contains several
<other SCSI chips and can  multiplex  one  SCSI address onto these SCSI chips
<(using the LUN) effectively extending the number of SCSI addresses available. 
<
That seems to be possible; you'd need a device with two SCSI ports.
It'd listen on both ports; requests which arrive on one port for LUN Z would be
translated to requests for SCSI ID Z on the other port, using LUN zero.
(Almost no SCSI device out there uses LUNs anyway.)
That way, you could connect up to 49 SCSI devices to one host.

The translation can be symmetric, with the SCSI ID on one port being
reserved as LUN for accepting commands (on the second port) for the device
itself. That way, you could do some configuring and maybe change the LUN used
on the other side, if absolutely necessary; if you do that, the number of
devices you can connect to your host becomes virtually unlimited (you can
daisy-chain the boxes) although you can't access all your disks at the same
time any more. ;-)

It seems that you'd need SCSI 2 with its capability to have multiple
outstanding commands per device if you want to access more than one device
through such a box at the same time. (The peripheral devices don't need
SCSI-2; only your host and the boxes need to.)

Hmmm... One 68000, two SCSI chips (NCR, whatever), a bit of EPROM and
some RAM, a power supply, and of course a month of time to build the hardware
and another to program the translation... Any takers?

-- 
Matthias Urlichs -- urlichs@smurf.sub.org -- urlichs@smurf.ira.uka.de     /(o\
Humboldtstrasse 7 - 7500 Karlsruhe 1 - FRG -- +49+721+621127(0700-2330)   \o)/