[ale] "Cost" of raid?

[Ryan Matteson]

I think the speed would depend on the implementation. When
I get new storage arrays or LVM releases, I tend to let
bonnie sort out the numbers for me. I usually test for:

- Small random I/O (Reads & Writes)
- Small sequential I/O (Reads & Writes)
- Large random I/O (Read & Writes)
- Large sequential I/O (Reads * Writes)

If RAID5 is used, I like to yank a disk and see how the
Hardware RAID controller/LVM packages handle reads and writes.

- Ryan

On Thu, 2002-09-26 at 23:40, Jeff Hubbs wrote:
> Just from what you're saying and presuming that you're going to do
> kernel RAID, You might want to lay off the RAID 5 if you're very
> write-dependent.  The kernel has to generate parity for writes, and if
> your machien is busy "doing some calculations" then you might feel some
> pain from that.  
> 
> If you can bring more drives to bear, however, you can split up your
> disk duties and set up some RAID 0+1 AND some RAID 5.
> 
> - Jeff
> 
> 
> On Thu, 2002-09-26 at 16:29, Robert L. Harris wrote:
> > 
> > 
> > Collecting data, doing some calculations on that data then pushing it
> > and the output to another machine.
> > 
> > Thus spake Jeff Hubbs (hbbs at attbi.com):
> > 
> > > From: Jeff Hubbs <hbbs at attbi.com>
> > > To: "Robert L. Harris" <Robert.L.Harris at rdlg.net>
> > > Cc: Atlanta Linux Enthusiasts <ale at ale.org>
> > > X-Mailer: Evolution/1.0.2-5mdk 
> > > Date: 26 Sep 2002 16:07:07 -0400
> > > Subject: Re: [ale] "Cost" of raid?
> > > 
> > > Robert -
> > > 
> > > What's the machine going to be doing?
> > > 
> > > - Jeff
> > > 
> > > 
> > > On Thu, 2002-09-26 at 15:58, Robert L. Harris wrote:
> > > > 
> > > > 
> > > > 
> > > >   I'm effectively "pricing" raid for a system on our network.  Currently
> > > > they have 4x18Gig disks, Raid 0 giving them 72Gigs available.  I've done
> > > > a breakdown of how much disk is available, what will be the effects on
> > > > space and redundance.  
> > > > 
> > > >   One question hit me I'm not sure how to answer though.  How much
> > > > slower is Raid 5 than a Raid 0+1?  Will it be noticable?  This system is
> > > > very IO intensive with an average load around 0.8.
> > > > 
> > > >   The box is a dual-P3-850 with 4x18Gig SCSI Cheetas.
> > > > 
> > > > 
> > > > :wq!
> > > > ---------------------------------------------------------------------------
> > > > Robert L. Harris                
> > > >                                
> > > > DISCLAIMER:
> > > >       These are MY OPINIONS ALONE.  I speak for no-one else.
> > > > FYI:
> > > >  perl -e 'print $i=pack(c5,(41*2),sqrt(7056),(unpack(c,H)-2),oct(115),10);'
> > > > 
> > > > 
> > > > ---
> > > > This message has been sent through the ALE general discussion list.
> > > > See http://www.ale.org/mailing-lists.shtml for more info. Problems should be 
> > > > sent to listmaster at ale dot org.
> > > > 
> > > 
> > > 
> > > 
> > > ---
> > > This message has been sent through the ALE general discussion list.
> > > See http://www.ale.org/mailing-lists.shtml for more info. Problems should be 
> > > sent to listmaster at ale dot org.
> > 
> > 
> > 
> > :wq!
> > ---------------------------------------------------------------------------
> > Robert L. Harris                
> >                                
> > DISCLAIMER:
> >       These are MY OPINIONS ALONE.  I speak for no-one else.
> > FYI:
> >  perl -e 'print $i=pack(c5,(41*2),sqrt(7056),(unpack(c,H)-2),oct(115),10);'
> > 
> 
> 
> 
> ---
> This message has been sent through the ALE general discussion list.
> See http://www.ale.org/mailing-lists.shtml for more info. Problems should be 
> sent to listmaster at ale dot org.
-- 
Ryan Matteson - UNIX Administrator
GPG ID: 782C2F4B Ryan Matteson (Matty) <matty91 at bellsouth.net>
Digitally Signed Public Key: http://www.daemons.net/~matty/public.txt
Fingerprint = C8D2 E7E8 815C 7AF6 59C4  C621 B829 A3D9 782C 2F4B

 This is a digitally signed message part