RE: Creation of ASM Disk Group on Solaris 10 ?
- From: "Randy Johnson" <randyjo@xxxxxxxxxxxxx>
- To: "'Alex Gorbachev'" <ag@xxxxxxxxxxxx>
- Date: Tue, 10 Jul 2007 17:28:33 -0500
Thanks Alex. I would point out that ASM Best Practices hold that you should
use a stripe size of 1M or as close as possible. If you venture away from
that then chances are you will impact performance overall.
"If using hardware RAID, make sure LUN stripe size is as close to
1mb as possible"
- "ASM Best Practices" by Nitin Vengurlekar
Truth is I'm still considering which is better. Seems like the premise of
improving performance using plaid is that you spread the physical read out
across multiple spindles. Perhaps a RAID 5 configuration is where plaid
would be more applicable. For example each LUN being a 4/8 disk RAID group.
I can see where hardware striping in additon to ASM striping might be a
little redundant although I've seen presentations that indicate it gives you
a favorable data spread for OLTP.
Thanks for your thoughts.
-Randy
Randy Johnson
Sr. Technical Consultant
Enkitec, LLP
Office ..... 817-255-3580
Mobile .... 817-564-6583
Email ..... randy.johnson@xxxxxxxxxxx
-----Original Message-----
From: gorbyx@xxxxxxxxx [mailto:gorbyx@xxxxxxxxx] On Behalf Of Alex Gorbachev
Sent: Tuesday, July 10, 2007 2:47 PM
To: randyjo@xxxxxxxxxxxxx
Cc: oracle-l@xxxxxxxxxxxxx
Subject: Re: Creation of ASM Disk Group on Solaris 10 ?
Once I've got in trouble when striping conflicted with each other.
It was for redo logs that are striped with fine striping in ASM (128K) and
this collided with stripe size of underlying volume so that one disk was
usually a bottleneck. But that's rather an exception.
I can imagine (and see it "on the paper") it's also possible to have
collision with other 1 MB stripe size but I haven't experienced that.
On the other hand, I should admit that statement "ASM works best if you can
avoid striped volumes" is *too* bold perhaps. :) I should have said
*usually* the best. One reason is that no stripe collision would be
possible. Another reason is that it's easy to provision space to disk group
(add/remove disks) providing that all disks are the same. With chunks taken
from the common RAID pool you have all chances to get one LUN from a more
loaded pool. Another "unfortunate" placement would be to get a chunk from
beginning of the disks in the pool and another chunks from the end - you
would have different performance from them but most important it would force
disk head jump over the whole disk back and force.
Also, I admit that there are arguably different stripe sizes that better fit
certain IO patterns and that might not fit within two standard 128K and 1M
stripe sizes so in some rare cases striping behind ASM might be appropriate.
I've personally seen that 64K stripe size was the best for one case where we
were able to compare it with all others on the real workload. However, this
brings additional things to keep in mind while provisioning the space and,
to some extent, defeats the main benefit of ASM - simplified storage
management.
A very good paper I can point to
http://www.oracle.com/technology/products/database/asm/pdf/take%20the%20gues
swork%20out%20of%20db%20tuning%2001-06.pdf
It has more details and, actually, benchmark there points to even slight
increase of performance with introduction of hardware striping.
But those are artificial tests and ideal layout used while in the real life
storage admins tend to abstract themselves and, of course, DBAs from
physical layout.
On 7/10/07, Randy Johnson <randyjo@xxxxxxxxxxxxx> wrote:
> Thanks Alex. Can you provide some documentation to support your "no
> SAN striping" point?
>
> -Randy
>
>
> Randy Johnson
> Sr. Technical Consultant
> Enkitec, LLP
>
> Office ..... 817-255-3580
> Mobile .... 817-564-6583
> Email ..... randy.johnson@xxxxxxxxxxx
>
>
> -----Original Message-----
> From: gorbyx@xxxxxxxxx [mailto:gorbyx@xxxxxxxxx] On Behalf Of Alex
> Gorbachev
> Sent: Monday, July 09, 2007 5:30 PM
> To: randyjo@xxxxxxxxxxxxx; oracle-l@xxxxxxxxxxxxx
> Subject: Re: Creation of ASM Disk Group on Solaris 10 ?
>
> Good answer for RTFM question Randy.
> Just a little comment about striping over striping. ASM works best if
> you can avoid striped volumes but, instead, can give separate physical
> spindle as the LUN. This is how it was designed to work.
> Unfortunately, in real life storage admins often can't be bothered.
> :-(
>
>
> On 7/9/07, Randy Johnson <randyjo@xxxxxxxxxxxxx> wrote:
> > Vivek,
> > This is the kind of question that if you have to ask you are not
> > ready for the answer. The best advice I could give you is to read
> > Nitin Vengurlekar's "ASM Best Practices" white paper and refer you
> > to Oracle's Cluster Ready ServicesInstallation Guide. If you've
> > never done this before you should brace yourself for some heavy reading.
> >
> > In short to answer your questions the raw devices (LUN's) you want
> > to use for ASM must be shared and visible to all nodes. This is
> > generally done using HBA (host bus adapter) cards that allow your
> > servers to "see and interact with" the SAN storage array. There are
> > some specifics in configuring ASM on Solaris (which slices you
> > can/should
> > use) and which slice to never use. Regarding commands for SAME. This
> > is done by having your storage administrator stripe the volumes
> > across
> several physical devices.
> > Then you as the DBA come through and configure your ASM volume
> > groups as a combination of these "striped" LUN's. In other words the
> > LUNS are striped on the SAN and the ASM Diskgroups provide a second
> > layer of
> striping.
> >
> > -Randy
> >
> >
> >
> > Randy Johnson
> > Sr. Technical Consultant
> > Enkitec, LLP
> >
> > Office ..... 817-255-3580
> > Mobile .... 817-564-6583
> > Email ..... HYPERLINK
> > "mailto:randy.johnson@xxxxxxxxxxx"randy.johnson@xxxxxxxxxxx
> >
> >
> >
> > _____
> >
> > From: oracle-l-bounce@xxxxxxxxxxxxx
> > [mailto:oracle-l-bounce@xxxxxxxxxxxxx]
> > On Behalf Of VIVEK_SHARMA
> > Sent: Monday, July 09, 2007 3:09 AM
> > To: oracle-l@xxxxxxxxxxxxx
> > Subject: Creation of ASM Disk Group on Solaris 10 ?
> >
> >
> >
> > Folks
> >
> >
> >
> > Need to Setup a 2 Instance ASM-RAC Database from scratch for an
> > internal Test Benchmark.
> >
> >
> >
> > * What Commands are used to create an External ASM Disk Group on
> > Solaris 10?
> >
> > * Any Additional Solaris Commands to make the SAME Disk Group
Visible
> > to BOTH SUN Nodes?
> >
> >
> >
> > NOTE - Disk Group to use External Hardware RAID (10) provided by the
> > Storage Box
> >
> >
> >
> > Oracle 10gR2 on Solaris 10
> >
> >
> >
> > Any Docs, Links will help
> >
> >
> >
> > Cheers & Thanks indeed
> >
> >
> >
> > Vivek
> >
> >
> >
> >
> >
> >
> >
> > **************** CAUTION - Disclaimer ***************** This e-mail
> > contains PRIVILEGED AND CONFIDENTIAL INFORMATION intended solely for
> > the use of the addressee(s). If you are not the intended recipient,
> > please notify the sender by e-mail and delete the original message.
> > Further, you are not to copy, disclose, or distribute this e-mail or
> > its contents to any other person and any such actions are unlawful.
> > This e-mail may contain viruses. Infosys has taken every reasonable
> > precaution to minimize this risk, but is not liable for any damage
> > you may sustain as a result of any virus in this e-mail. You should
> > carry out your own virus checks before opening the e-mail or attachment.
> > Infosys reserves the right to monitor and review the content of all
> messages sent to or from this e-mail address.
> > Messages sent to or from this e-mail address may be stored on the
> > Infosys e-mail system.
> > ***INFOSYS******** End of Disclaimer ********INFOSYS***
> >
> >
> >
> > No virus found in this incoming message.
> > Checked by AVG Free Edition.
> > Version: 7.5.476 / Virus Database: 269.10.2/891 - Release Date:
> > 7/8/2007
> > 6:32 PM
> >
> >
> >
> > No virus found in this outgoing message.
> > Checked by AVG Free Edition.
> > Version: 7.5.476 / Virus Database: 269.10.2/891 - Release Date:
> > 7/8/2007
> > 6:32 PM
> >
> >
>
>
> --
> Alex Gorbachev, Oracle DBA Brewer, The Pythian Group
> http://www.pythian.com/blogs/author/alex http://blog.oracloid.com BAAG
> party
> - www.BattleAgainstAnyGuess.com
>
> No virus found in this incoming message.
> Checked by AVG Free Edition.
> Version: 7.5.476 / Virus Database: 269.10.2/891 - Release Date:
> 7/8/2007
> 6:32 PM
>
>
> No virus found in this outgoing message.
> Checked by AVG Free Edition.
> Version: 7.5.476 / Virus Database: 269.10.2/891 - Release Date:
> 7/8/2007
> 6:32 PM
>
>
>
--
Alex Gorbachev, Oracle DBA Brewer, The Pythian Group
http://www.pythian.com/blogs/author/alex http://www.oracloid.com BAAG party
- www.BattleAgainstAnyGuess.com
No virus found in this incoming message.
Checked by AVG Free Edition.
Version: 7.5.476 / Virus Database: 269.10.2/893 - Release Date: 7/9/2007
5:22 PM
No virus found in this outgoing message.
Checked by AVG Free Edition.
Version: 7.5.476 / Virus Database: 269.10.2/893 - Release Date: 7/9/2007
5:22 PM
--
//www.freelists.org/webpage/oracle-l
Other related posts:
