Re: 3 X 18 GB disks

From: <>
Date: Mon, 30 Jun 2008 23:52:52 -0700 (PDT)
Message-ID: <>

Arch wrote:

> On Mon, 30 Jun 2008 16:55:55 -0700, DA Morgan <>
> wrote:
> > wrote:
> >> Hello,
> >>
> >> I got some guidance earlier (TA!) about what possibilities their were
> >> for slicig and dicing 3 X18 GB disks on a Sun 440 development box for
> >> Oracle 10G. So now current plan is to put Solaris 10 and query app on
> >> disk1, oracle product on a 5G B slice /u01 on disk2 and data on disk3 /
> >> u02.
> >> I know the data is not secure cos nothing is mirrored, but we are
> >> going to run in for a month, which should produce about 18GB of data,
> >> before we can move to production.
> >>
> >> Maybe I should create another slice or two on disk3 and use them for
> >> logging for example, to try and add some performance - I have not yet
> >> installed 10G R2 and am not shooting for ASM this time round.
> >>
> >> Any hints appreciated!
> >>
> >> Max
> >
> >My personal recommendation would be to use them as a boat anchor and
> >get one really fast SATA drive for $50.
> >
> >But if they are what you have ... then SAME ... Stripe and Mirror
> >Everything.
> Max,
> I use a 440 for development - does a fine job. If you really are
> limited to 3 18GB disks, however, you don't have enough drives or disk
> space to stripe and mirror. Slicing the disks up the way you are
> describing is a waste of time. You will get zero benefit and quite
> probably make it more difficult to manage the space for your database.
> When you install Solaris, create the least number and largest size
> partitions that you can.
> BTW - SCSI drives are cheap on eBay.
> Good Luck,
> Arch

Good points about upgrading the number and type of disks. I dunno about the least number of partitions, This is a dev box and will move to production after it has been proved it works, I actually read the install instructions for 10G on Sol10 ans in the appendix it suggests mount points /u01 and /u02 - I thoght there was some benefit in having mountpoints mapping the slices so that if we really did run out of space a new disk or two could replace a slice. Cos i reckon that actually running out of space might prompt the purchase of extra disks, and i want the dev architecture to be scaleable (ie just add hardware). Received on Tue Jul 01 2008 - 01:52:52 CDT

Original text of this message