Re: ASM and disk partitions offset
- From: "Radoulov, Dimitre" <cichomitiko@xxxxxxxxx>
- To: Niall Litchfield <niall.litchfield@xxxxxxxxx>
- Date: Thu, 16 Feb 2012 11:53:19 +0100
Hi Niall,
thank you very much.
The missing bit (for me) was:
> most vendors recommend the 1mb starting point for a partition because
> it has as a common divisor most of the usual allocation unit sizes and disk
> label size
We have already the disk groups for crs (ocr and voting), I may try to
play with those too - move the voting and the ocr)
But I'll definitely try to talk again to the storage team in order to
partition the LUNs for the database disk groups using the correct
offset.
Thanks again!
Best regards
Dimitre
On Thu, Feb 16, 2012 at 11:34 AM, Niall Litchfield
<niall.litchfield@xxxxxxxxx> wrote:
> Dimitre,
>
> "I wouldn't bother" is probably incorrect. The details will depend on the
> array stripe size, the sector sizes and probably the specific hardware. That
> said most vendors recommend the 1mb starting point for a partition because
> it has as a common divisor most of the usual allocation unit sizes and disk
> label sizes. I quite like this article
> http://blogs.msdn.com/b/jimmymay/archive/2009/05/08/disk-partition-alignment-sector-alignment-make-the-case-with-this-template.aspx
> written from a windows perspective with SQL in mind, but really the basic
> issue isn't software specific. FWIW Oracle, Microsoft and VMWare all
> recommend that you align your storage hardware with the FS/Volume manager so
> as to avoid doing multiple IO requests where a single request would do,
> references below.
>
> http://www.vmware.com/pdf/Perf_Best_Practices_vSphere4.0.pdf
> http://www.oracle.com/technetwork/articles/systems-hardware-architecture/lun-alignment-163801.pdf
> http://support.microsoft.com/kb/929491
>
>
>
>
>
>
> On Thu, Feb 16, 2012 at 10:12 AM, Radoulov, Dimitre <cichomitiko@xxxxxxxxx>
> wrote:
>>
>> On Thu, Feb 16, 2012 at 7:29 AM, Stalin Subbiah <stalinsk@xxxxxxxxx>
>> wrote:
>> > I believe it was also done to align ASM disks with stripe size which is
>> > generally in multiples of 2. At least this was the case with Linux, using
>> > fdisk for disk partitions.
>>
>> Thanks Stalin,
>> but what that means: that a partition beginning at the first cylinder
>> is not aligned with the HW stripe size (presumably a power of 2)?
>>
>> I've asked if the storage team had some special requirement (or
>> suggestion) as far as the first cylinder of the partitions is
>> concerned and I've received a generic reply that I shouldn't bother.
>>
>>
>> Best regards
>> Dimitre
>>
>>
>> > On Feb 13, 2012, at 3:13 PM, "Radoulov, Dimitre" <cichomitiko@xxxxxxxxx>
>> > wrote:
>> >
>> >> I meant:
>> >>
>> >> now I remember that once, when I was setting up a test RAC on Solaris,
>> >> I
>> >> wasn't able to make the partitions visible for asm disk creation
>> >> [...]
>> >> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>> >>
>> --
>> //www.freelists.org/webpage/oracle-l
>>
>>
>
>
>
> --
> Niall Litchfield
> Oracle DBA
> http://www.orawin.info
--
//www.freelists.org/webpage/oracle-l
Other related posts:
