From owner-freebsd-questions  Wed Apr 17 22:43:33 1996
Return-Path: owner-questions
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.3/8.7.3) id WAA15627
          for questions-outgoing; Wed, 17 Apr 1996 22:43:33 -0700 (PDT)
Received: from shell.aros.net (shell.aros.net [205.164.111.19])
          by freefall.freebsd.org (8.7.3/8.7.3) with ESMTP id WAA15622
          Wed, 17 Apr 1996 22:43:31 -0700 (PDT)
Received: (from angio@localhost) by shell.aros.net (8.7.5/Unknown) id XAA15760; Wed, 17 Apr 1996 23:43:25 -0600 (MDT)
From: Dave Andersen <angio@aros.net>
Message-Id: <199604180543.XAA15760@shell.aros.net>
Subject: Re: SCSI RAID controller support?
To: freebsd-questions@freebsd.org
Date: Wed, 17 Apr 1996 23:43:24 -0600 (MDT)
Cc: gpalmer@freebsd.org, thekind@NETural.com
In-Reply-To: <199604172259.PAA02918@phaeton.artisoft.com> from Terry Lambert at "Apr 17, 96 03:59:37 pm"
X-Mailer: ELM [version 2.4ME+ PL13 (25)]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: owner-questions@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

Lo and behold, Terry Lambert once said:
> > I'm surprised that you need RAID for web serving at all. You'd need a
> > VERY high hit rate, or be pumping out large documents to need such
> > access speed.
> 
> What?  What makes you think RAID is faster?  It's slower, without
> hardware acceleration (like an NVRAM write cache).  You have to do
> two writes for each write, otherwise...
> 
> If he wants addes speed, he should use striping with spindel-sync,
> not RAID.
> 
> RAID is for fault tolerance and error recovery.

   That depends entirely on the level of RAID you're using.  You can use 
RAID for fault tolerance, -or- for disk striping, or both.

   RAID 0:  Striping, no parity.
            Not true "raid" but often sold as it.
   RAID 1:  Mirroring on two disks - redundancy - same speed
            as a normal drive (in theory. :)  The ECC may slow things down.
   RAID 2:  Hamming ECC - basically for data redundancy
            About as fast as raid 1, perhaps a bit faster.
   RAID 3:  Striping with parity checking
            A bit more reliable than RAID 0, not quite as fast,
            but still considerably faster than a straight disk
   RAID 4:  Parity Checking on a special parity disk.
   RAID 5:  Parity checking with parity distributed across data disks

  Quite obviously, RAID 0 and RAID 3 have the potential to be 
considerably faster than an ordinary disk.  I don't know how they compare 
to software striping as in the ccd, I don't think anyone's done any 
comparisons.:)  I've used some of these in graphic design applications, 
and they *really* fly.  A RAID level 0 array like the FWB Jackhammer is a 
very pretty piece of equipment, though a tad expensive for most people. :)

  Granted that RAID 0 isn't really "R"aid in the sense that it isn't 
redundant, but even RAID 3 is faster than a standalone disk (and has the 
advantage of redundancy), and many manufacturers sell striped 
non-redundant disk arrays as RAID drives.

     -Dave Andersen

-- 
angio@aros.net                Complete virtual hosting and business-oriented
system administration         Internet services.  (WWW, FTP, email)
http://www.aros.net/          http://www.aros.net/about/virtual
  "There are only two industries that refer to thier customers as 'users'."