Re: ASM+internal disks

From: Greg Rahn <>
Date: Tue, 26 Apr 2011 10:52:08 -0700
Message-ID: <>

Seems to me that all of these bugs have been fixed already as of RHEL 5.5. and 5.6 is already out.

For the OP:
+1 for what Kevin says - just HW raid all 8 drives and skip ASM. If this system is going to be pushed hard, just use XFS with the deadline scheduler for the database datafile volume(s). ASM offers very little on such a small installation.

On Tue, Apr 26, 2011 at 9:16 AM, Vishal Gupta <> wrote:
> On RHEL5, CFQ i/o scheduler is broken. Under heavy load performance improves
> with deadline i/o scheduler. On RHEL4 it does not give much benefit. There
> is linux bug for broken CFQ on RHEL5.
> Regards,
> Vishal Gupta
> ________________________________
> From: on behalf of Kevin Closson
> Sent: Tue 26/04/2011 16:45
> To: Oracle-L
> Subject: Re: ASM+internal disks
> Here's how I look at this. The following are true 1) there is no networked
> storage so all disks are the same, 2) the database is important and 3) the
> Oracle software and OS are equally important. Also, in spite of the fact
> that I got no information back regarding how the disks are attached (e.g.,
> the make and model of the RAID card) I think we can all be quite certain
> that the card supports HW raid. To that end, I would simply configure all 8
> drives in a HW raid configuration. I see no benefit in the slice and dice. I
> also see no value in ASM in such a situation but I suppose this could be
> standard edition or some other such software that forces its use?
> So, Niall, we agree: " RAID? Seriously?"  :-)

Greg Rahn
Received on Tue Apr 26 2011 - 12:52:08 CDT

Original text of this message