zfs discuss - Nov 2009 - ZFS and oracle on SAN disks

I believe the best practice is to use seperate disks/zpool for oracle database
files as the record size needs to be set the same as the db block size - when
using a jbod or internal disks.

If the server is using a large SAN LUN can anybody see any issues if there is
only one zpool and the dataset is set to have the record size set rather than at
the zpool level?

Thanks

Ian
-- 
This message posted from opensolaris.org

On Nov 10, 2009, at 5:32 AM, Ian Garbutt wrote:
> I believe the best practice is to use seperate disks/zpool for  
> oracle database files as the record size needs to be set the same as  
> the db block size - when using a jbod or internal disks.
recordsize is not a pool property, it is a dataset (zvol or file  
system) property.
You can have different recordsize or volblocksize property settings  
for each
dataset in the pool.  You can also have different cache policies,  
which will
also be useful. Roch reports a 40% performance improvement by adjusting
the logbias property.
	http://blogs.sun.com/roch/entry/synchronous_write_bias_property

However, in that case, a "large SAN LUN" was a hybrid storage pool.
If you are using a traditional RAID array, then the nonvolatile write  
cache
can also improve performance.
  -- richard
>
> If the server is using a large SAN LUN can anybody see any issues if  
> there is only one zpool and the dataset is set to have the record  
> size set rather than at the zpool level?
>
> Thanks
>
> Ian
> -- 
> This message posted from opensolaris.org
> _______________________________________________
> zfs-discuss mailing list
> zfs-discuss at opensolaris.org
> http://mail.opensolaris.org/mailman/listinfo/zfs-discuss