Re: RAID 0, 1, 5

Tech-Archive recommends: Fix windows errors by optimizing your registry

From: John Harris (x_at_x.x)
Date: 02/19/04


Date: Thu, 19 Feb 2004 21:07:03 +1300

Hi Dave and IBC
I noticed you thread and thought I might add a little to it. I run RAID1
for the system drive and RAID5 with a hot spare for the data. The reasons
for this are the RAID1 allows me to remove a drive before hot patching and
then ensuring everything still works after the patch. If not, use the prior
removed drive to take you back (an idea I stole from someone else). RAID5
for the data drive for the reasons Dave has espoused previously.

Harry

"Dave Nickason [SBS MVP]" <gwdibble@NOSPAM.frontiernet.net> wrote in message
news:%23ZYEHvl9DHA.340@tk2msftngp13.phx.gbl...
> We actually have a PE2400, a 2500 (SBS) and a 2600 (document server). The
> 2400 does printer sharing, accounting, and phone call logging, all of
which
> could be done on the SBS. The 2400 was the document server, but we needed
> to upgrade that, which is why we ended up with three boxes instead of two.
> All three have hot spare drives, which is a pretty big investment in
total,
> but we defnintely want them on the SBS and the document server. You
always
> hear that it's not if but when for hard drive failures, but I've never had
> one until this recent incident. Dell does a fair amount of bragging about
> their raid controllers, and it's nice to see that at least some of it is
> true!
>
>
>
>
> "IBC" <spamityspam@spam.spam> wrote in message
> news:OTkfpYl9DHA.3880@TK2MSFTNGP11.phx.gbl...
> > I remembered you talked about your set up before Dave, and I was hoping
> you
> > would comment on your reasons. Thank you. I would THINK that on ANY SBS,
> the
> > entire box is mission critical. We only run one server here in our
office
> > (oh boy is that going to change though....) and if we didn't have RAID5
> and
> > Hot Swapability, well, to quote the new FedEx commercials, we would be
> > "DOOMED!!" We were only 12-15 people then, so the performance hit wasn't
> the
> > end of the world, but now that we've grown some perhaps I should
consider
> > this in our next server. What do Best Practices say?
> >
> > On an interesting side note to match Dave's, The drive that failed on us
> was
> > from a Dell PE2400 an it failed within 2 months of being installed.
> > Troubleshooting was as quick and the turn around on the part was
> > outstanding. (Failed late in the afternoon, part was sitting at the door
> > when I showed up the next morning.)
> >
> >
> > "Dave Nickason [SBS MVP]" <gwdibble@NOSPAM.frontiernet.net> wrote in
> message
> > news:#DHwM#k9DHA.2696@TK2MSFTNGP10.phx.gbl...
> > > I have a new Dell PE2600 with 5 36g drives, 4 in a RAID 5 and one
> > dedicated
> > > spare. This server was built in Nov. and I just put it into
production
> in
> > > late Dec. of last year. Two weeks ago on a Sunday, a drive failed
> without
> > > warning. I didn't notice it for three days because the controller
> > > automatically and transparently failed over to the spare drive. The
> only
> > > server down time was when Dell requested me to reboot into the
> controller
> > > config software to verify that there were no backplane errors logged
and
> > > that the problem was in fact the drive. I installed the new drive
hot,
> > and
> > > configured it as the new spare - no impact to the users whatsoever.
> > >
> > > If two drives fail simultaneously, the spare won't help. If two fail
> > > sequentially, you can run in the configuration you describe until you
> can
> > > replace at least one of the drives. The Dell tech said that most
people
> > > don't run with spares because of the cost, but to us, the extra
> redundancy
> > > and the lack of user impact are worth it. That server runs our
document
> > > repository housing all of our case files, and if it's down, a lot of
> > people
> > > are left with nothing to do.
> > >
> > > A Dell comment - I talked to tech support about the failed drive at
6:30
> > pm.
> > > Total time to verify that it was the drive and not the backplane, and
to
> > > arrange for a new drive: 15 minutes. They offered to have someone
over
> to
> > > replace it that night, but I declined since I had somewhere else to
be.
> > > When I got here the next morning, a new drive was waiting for me.
> > >
> > >
> > > "IBC" <spamityspam@spam.spam> wrote in message
> > > news:uLferQk9DHA.2308@TK2MSFTNGP11.phx.gbl...
> > > > I still don't get the hot swap drive idea. It is nice, but you have
> > > > effectively taken 4 drives and made 2 available. 1 sits out as a
> spare,
> > > and
> > > > the other is 'lost' to the RAID5 config. Am I nutzo here? What I
> recall
> > > is
> > > > that the capacity was always the total of all drives minus the one
> that
> > > was
> > > > consumed for RAID 'workings' for lack of a better term. (It think
its
> > > > actually called Parity, but I digress....)So if you have 5 25 GB HDs
> in
> > > > RAID5, you only get 100GB of usable storage. If you had 3, You would
> > only
> > > > get 50GB. The beauty of RAID5 is that you can lose an entire drive
and
> > the
> > > > system keeps going. We lost a drive once and it was out for 36
hours.
> We
> > > > never lost a step, and when the new drive came we slapped it in, the
> > RAID
> > > > rebuilt invisibly in the background and on we went. I'd rather not
> risk
> > > > running out of space or have to spend the time reconfiguring the
RAID
> > down
> > > > the road.
> > > >
> > > > Now, I suppose there is the possibility of more than one drive
failing
> > at
> > > > the same time, but in that case 1 spare isn't going to help anyhow,
is
> > it?
> > > >
> > > > As always, I may be way off base here, so you may want more input
from
> > > > somebody with more letters behind their name. =P
> > > >
> > > >
> > > >
> > > >
> > > > "Marcia Porter" <mporter@shpam.acbfparentadvocates.org> wrote in
> message
> > > > news:uUYJS8j9DHA.2028@TK2MSFTNGP10.phx.gbl...
> > > > > Hi! We just purchased a new server that has 4 hard drives that
can
> > > handle
> > > > > the following combinations:
> > > > > 0, 1, 5, 0/1, 0/5. Which do most SBSers do? The server will be
the
> > > > > Exchange, ISA, and SQL server.
> > > > >
> > > > > Thanks.
> > > > >
> > > > > Marcia
> > > > >
> > > > >
> > > >
> > > >
> > >
> > >
> >
> >
>
>



Relevant Pages

  • Re: RAID 0, 1, 5
    ... All three have hot spare drives, which is a pretty big investment in total, ... but we defnintely want them on the SBS and the document server. ...
    (microsoft.public.backoffice.smallbiz2000)
  • Free MicroVAXen in Central Illinois.
    ... I have 3 q-bus MicroVAXen + some spare parts that I need to get rid of. ... Inside are SCSI drives and RD54 drives. ... I suspect that one of the SCSI disks failed in a way that is causing ... installation on an RD54 and then boot it on the MicroVAX I, ...
    (comp.os.vms)
  • Re: DELL MD1000 for SBS2003???
    ... OS, RAID1 for the OS, RAID5 for the data, partitioned however your ... so the hot spare is a great help. ... with 8 drives, and the only reason I upgraded from 6 drives was because ... word docs OR 2 partitions 1 for each use??? ...
    (microsoft.public.windows.server.sbs)
  • Re: mdadm Raid Management - Assigning Hot Spares
    ... >> the spare devices. ... DS14 14*36GB drives array available for Linux. ... st -F /dev/sg1 to low level format. ...
    (comp.os.linux.misc)
  • Re: SBS 2003 Partition Questions
    ... I mirror the OS on one channel and srtipe the ... rest using a hot spare. ... your mind that their "has" to be a performance difference and that is good ... > I'm pretty sure holds 8 drives. ...
    (microsoft.public.windows.server.sbs)