[1399] in linux-scsi channel archive
Re: ncr53c8xx problem with a CLARiiON disk array.
daemon@ATHENA.MIT.EDU (Dario_Ballabio@milano.europe.dg.co)
Fri Feb 7 03:24:09 1997
Date: Fri, 7 Feb 1997 09:22:30 +0100
From: Dario_Ballabio@milano.europe.dg.com
To: linux-scsi@vger.rutgers.edu, groudier@club-internet.fr
Your workaround is not feasible. The CLARiiON disk array has one or
two storage processors. Each storage processor has 1 fast wide diff
scsi controller to the host and internally 5 wide scsi busses to the disks.
Internally there are 20 drives, so each internal scsi bus takes 4
drives. The storage processor has a single target id on the diff bus
seen by the host ncr adapter, while each lun is seen by the host as a
single physical disk, but it can be either a RAID0, RAID1, RAID3 or
RAID5 with a number of drives (from 2 to 15). The classical setup is
4 RAID5 stripes each of 5 drives, so each drive in a stripe is on
a different internal scsi bus. In this way if a scsi bus fails,
a single drive in each stripe fails, bat the 4 RAID5 discs continue
to work normally. There is no other way to acces the physical units
than using its lun, while it is impossible to see each separate
drive on the physical unit.
For more info about this hardware, see http://www.clariion.com
and http://www.dg.com.
- db
