Re: Filesystem blocksize for archive logs on AIX
From: Nuno Souto <dbvision_at_iinet.net.au>
Date: Tue, 29 Oct 2013 20:18:21 +1100
Message-ID: <526F7D5D.30005_at_iinet.net.au>
On 29/10/2013 3:00 AM, Rich Jesse wrote: > So, a 4K agblksize and 8K DB_BLOCK_SIZE?
> Interesting! I wonder how the VMAX compares with the XIV for performance.
Date: Tue, 29 Oct 2013 20:18:21 +1100
Message-ID: <526F7D5D.30005_at_iinet.net.au>
On 29/10/2013 3:00 AM, Rich Jesse wrote: > So, a 4K agblksize and 8K DB_BLOCK_SIZE?
I'm not sure what you mean by "a 4K agblocksize".
JFS2 allows only a max 4K file system blocksize.
That doesn't mean by any stretch the db cannot use larger block sizes.
What does destroy the performance is to use a file system block size
larger than the db block size.
No danger of that with JFS2 and db_block_size=8K! :)
But a clear danger of that, if using std redos (512) and a default file
system blocksize of 4K,
as the other poster pointed out. That's when one has to match things.
One way or the other.
>> We use a VMAX SAN with 4Gbps FC cards for P6 and 8Gbps FC cards for >> the P7.
> Interesting! I wonder how the VMAX compares with the XIV for performance.
Sorry, I don't have any comp figures. FWIW, we use also EMC's Powerpath. Which greatly enhances the performance for JFS2 and Aix.
> How many members in your redolog groups?
See the other reply.
-- Cheers Nuno Souto dbvision_at_iinet.net.au -- http://www.freelists.org/webpage/oracle-lReceived on Tue Oct 29 2013 - 10:18:21 CET