Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!sdd.hp.com!spool.mu.edu!snorkelwacker.mit.edu!bloom-beacon!eru!kth.se!sunic!dkuug!ncrsecp!ncrlnk!ncrwic!encad!enint!hburford
From: hburford@enint.Wichita.NCR.COM (Harry Burford)
Newsgroups: comp.periphs.scsi
Subject: Re: RAID in hardware?
Message-ID: <748@enint.Wichita.NCR.COM>
Date: 27 Feb 91 14:07:38 GMT
References: <540@zds-ux.UUCP> <3229@legs.UUCP>
Organization: NCR Corporation, Wichita, KS
Lines: 27

cgn@ast.COM (Chris Nieves) writes:

|>In article <540@zds-ux.UUCP>, bjstaff@zds-ux.UUCP (Brad Staff) writes:
|>> Is anybody working on implementing RAID algorithms in hardware?  I'd

|>It doesn't make a whole lot of sense to have your cpu reading data,
|>XORing it, and writing the parity to memory, BEFORE you even start
|>the disk request.  It seems best to have a SCSI (IDE?) controller
|>handling all of the parity generation and data correction.  This way
|>the OS would need to have very little knowledge that there is anything
|>special out there other than a real big, fast disk.
|>Chris Nieves
|>AST Research
|>Anyone thought about starting a new group: comp.periphs.raid???

Very interesting!  The NCR Array controller handles all the data
correction in hardware so the operating system thinks its talking
to a real big, fast, SCSI disk.  Doing RAID in software DOES make
sense, however, if your major goal with ARRAY use is data 
reliability.  Cost for special hardware is reduced, but you will pay
a system performance penalty.

-- 
Harry Burford - NCR Peripheral Products Division       CALL:KA0TTY
SLOWNET:3718 N. Rock Road, Wichita KS 67226-1397    CI$:76226,2760
PHONE:316-636-8016  TELEX:417-465  FAX:316-636-8086         SS:9.5
Harry.Burford@Wichita.NCR.COM                'ask me about ARRAYS'