[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference ssag::ask_ssag

Title:	Ask the Storage Architecture Group
Notice:	Check out our web page at http://www-starch.shr.dec.com
Moderator:	SSAG::TERZAN

Created:	Wed Oct 15 1986
Last Modified:	Fri Jun 06 1997
Last Successful Update:	Fri Jun 06 1997
Number of topics:	6756
Total number of notes:	25276

6603.0. "A4100, DU3.2g & Boxhill Raid Controller" by SIOG::BR_MURPHY () Mon Apr 21 1997 06:51

    This note is crossposted on the Digital Unix Conference 9564
    
    Has anyone out there had anything to do with Boxhill raid controllers?
    
    I have a customer who has a Boxhill raid controller. Over the weekend
    the VAR who supplied this setup a number of raidsets. However this
    controller appears to transmit on SCSI id 0, LUN 0,1,2...ad infinitum.
    Actually they created 17 LUNS. The first group of raidsets delivering 7
    lots of 5Gb disk space on SCSI ID 0, LUNS 0-6 worked fine. The next
    group of raidsets delivering 7 lots 5Gb of disk space on SCSI ID 0,
    LUNS 7-13 failed. The third group of raidsets delivering 3 lots of 5Gb
    disk space on SCSI ID 0, LUNS 14-16 also failed
    
    They say using multiple LUNS is normal & has been implemented on a
    number of different unix platforms successfully. I say this is not
    possible, due to hardware restrictions. The SCSI card (KZPSA) is only
    capable of supporting 8 SCSI target ID's, & for each target ID, 8 LUN
    numbers.
    
    They also said the Boxhill raid controller could only deliver raidsets
    via one SCSI target ID....unlike an HSZ device that can use multiple
    SCSI targets & luns.
    
    In the end, the raidsets were delivered large 35Gb disks & partitioned
    into 5Gb filesystems using disklabel. I'm not happy with this solution,
    as any changes to the size on one filesystem, by adding more disks to
    the controller affects every other filesystem...ie 35 GB will need to
    be backed up, the disklabel re-editted & all the data restored. This is
    not a satisfactory solution at all.
    
    Can anyone help.
    
    Brendan

T.R Title User Personal
Name Date Lines

6603.1 StorageWorks fixes this NQOS01::nqsrv508.nqo.dec.com::smith_p Wed Apr 23 1997 11:24 6

T.R	Title	User	Personal Name	Date	Lines
6603.1	StorageWorks fixes this	NQOS01::nqsrv508.nqo.dec.com::smith_p		`Wed Apr 23 1997 11:24`	6
	You are absolutely correct on your assumtions. SCSI specs state that each SCSI target can support up to 8 LUN's. To add additional LUN's additional SCSI targets must be added. If only one target exists, o-well, they should have bought StorageWorks and this would not happen. Paul

You are absolutely correct on your assumtions.  SCSI specs state that each 
SCSI target can support up to 8 LUN's.  To add additional LUN's additional 
SCSI targets must be added.  If only one target exists, o-well, they should 
have bought StorageWorks and this would not happen.

Paul