Prev: select_catalog_role does not let me see data dictionary . why?
Next: Collection
From: DG problem on 23 Jun 2010 01:39
Running 9.2.0.8 (still) on HP-UX 11.11
2TB database Used for OLTP and reporting
New SAN is a Hitachi 5TB AMS2100

It seems that these days the simple guidelines for configuring a large
Oracle database on a SAN is to simply configure the SAN with a one
megabyte stripe width across all disks (2 x 8 disks plus one spare).
And, to use a cache block size which is the same size as the Oracle
DB_BLOCK_SIZE which is 8192 (8kB)?

All redo logs, archive logs, datafiles and tempfiles would be on the
same mountpoint across the 1MB stripe.

We also plan to use RAID 1+0

Is this configuration in the right ball park?

Any suggestions are welcome.


I took the 1MB stripe width size from the following document.
Deploying, Managing, and Administering the Oracle Internet Platform
Optimal Storage Configuration Made Easy
Juan Loaiza, Oracle Corporation

Which states:
1) Stripe all files across all disks using a one megabyte stripe
width.
2) Mirror Data for high availability.
From: Frank van Bortel on 23 Jun 2010 05:00
On 06/23/2010 07:39 AM, DG problem wrote:
> Running 9.2.0.8 (still) on HP-UX 11.11
> 2TB database Used for OLTP and reporting
> New SAN is a Hitachi 5TB AMS2100
>
> It seems that these days the simple guidelines for configuring a large
> Oracle database on a SAN is to simply configure the SAN with a one
> megabyte stripe width across all disks (2 x 8 disks plus one spare).
> And, to use a cache block size which is the same size as the Oracle
> DB_BLOCK_SIZE which is 8192 (8kB)?
>
> All redo logs, archive logs, datafiles and tempfiles would be on the
> same mountpoint across the 1MB stripe.
>
> We also plan to use RAID 1+0
>
> Is this configuration in the right ball park?
>
> Any suggestions are welcome.
>
>
> I took the 1MB stripe width size from the following document.
> Deploying, Managing, and Administering the Oracle Internet Platform
> Optimal Storage Configuration Made Easy
> Juan Loaiza, Oracle Corporation
>
> Which states:
> 1) Stripe all files across all disks using a one megabyte stripe
> width.
> 2) Mirror Data for high availability.

Make it all SAME: Stripe and Mirror Everything.
Don't know about stripe sizes - what do you think to gain by
playing with these? Did you look at the release date
of aforementioned document? If you want to deviate from
the vendors recommendations of stripe sizes, make sure
to test, Test and TEST.

If anything else than vendor recommendation comes
from that, the ng would appreciate feedback. Did
I mention testing?

--

Regards,

Frank van Bortel (who's a member of BAARF)
From: Mladen Gogala on 23 Jun 2010 10:07
On Wed, 23 Jun 2010 11:00:11 +0200, Frank van Bortel wrote:

> Make it all SAME: Stripe and Mirror Everything.

You're in marketing? SAME == RAID 1+0.

> Don't know about stripe sizes -

That was the OP's question.

> what do you think to gain by playing with these?

Speed.

And yes, stripe size of 1MB is good. Allegedly, 11G will work well with
4M, but I haven't had an opportunity to try that.




--
http://mgogala.byethost5.com
From: DG problem on 24 Jun 2010 02:47
On Jun 24, 12:07 am, Mladen Gogala <gogala.mla...(a)gmail.com> wrote:
> On Wed, 23 Jun 2010 11:00:11 +0200, Frank van Bortel wrote:
> > Make it all SAME: Stripe and Mirror Everything.
>
> You're in marketing? SAME == RAID 1+0.
>
> > Don't know about stripe sizes -
>
> That was the OP's question.
>
> > what do you think to gain by playing with these?
>
> Speed.
>
> And yes, stripe size of 1MB is good. Allegedly, 11G will work well with
> 4M, but I haven't had an opportunity to try that.
>
> --http://mgogala.byethost5.com

Just some more notes to confirm that I have this correct.


Stripe depth is the size of the stripe, sometimes called stripe unit.
Stripe width is the product of the stripe depth and the number of
drives in the striped set.
http://download.oracle.com/docs/cd/B28359_01/server.111/b28274/iodesign.htm#i28412

segment size = stripe depth
1MB stripe width = eight drives * stripe depth = 8 * 128KB

This SAN parameter is dynamic apparently (guess it depends on brand)
SAN Cache block size = SAN cache page size = DB_BLOCK_SIZE = 8KB
DB_BLOCK_SIZE should = file block size which should = cache block
size
http://www.emc.com/collateral/hardware/white-papers/h796-implementing-oracle-emc-clariion-stor-sys-wp.pdf

We probably don't need to worry about making the log files non-cached
as the cache is battery backed up?
From: John Hurley on 24 Jun 2010 09:23
DG:

> Just some more notes to confirm that I have this correct.
>
> Stripe depth is the size of the stripe, sometimes called stripe unit.
> Stripe width is the product of the stripe depth and the number of
> drives in the striped set.http://download.oracle.com/docs/cd/B28359_01/server.111/b28274/iodesi...
>
> segment size = stripe depth
> 1MB stripe width = eight drives * stripe depth = 8 * 128KB
>
> This SAN parameter is dynamic apparently (guess it depends on brand)
> SAN Cache block size = SAN cache page size = DB_BLOCK_SIZE = 8KB
> DB_BLOCK_SIZE should = file block size which should = cache block
> sizehttp://www.emc.com/collateral/hardware/white-papers/h796-implementing...
>
> We probably don't need to worry about making the log files non-cached
> as the cache is battery backed up?

Your messed up if you can't depend on the cache of your storage array.

Most of my experience with Oracle systems has been setting up OLTP
( not big datawarehouse ) systems.

I almost always separate out the disks for online logs from the rest
of the system. For commit happy systems I really want to have average
log file sync times at 1 ms or under if I can.

These days my systems are using ASM but I still setup separate
diskgroups for the online log files and separate the disks and that
stuff out.
 |  Next  |  Last
Pages: 1 2
Prev: select_catalog_role does not let me see data dictionary . why?
Next: Collection