Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Mailing Lists -> Oracle-L -> Re: Disk Async IO

Re: Disk Async IO

From: Nuno Souto <dbvision_at_iinet.net.au>
Date: Thu, 07 Sep 2006 20:29:58 +1000
Message-ID: <44FFF4A6.6010807@iinet.net.au>


BN wrote,on my timestamp of 7/09/2006 4:54 AM:

> I remember reading Oracle Metalink Notes saying that Disk Async IO is
> not supported on HP-UX for Cooked Filesystems.

er...
you're correct in the Metalink references. But there is no such thing as "cooked filesystems".
There is "raw" IO and "cooked" IO: the first is *without* filesystems, the second *with* filesystems.

> We have disk_async_io=true for our Database in HP-UX 11.11 , Oracle
> 8.1.7.4 <http://8.1.7.4> and 10g Servers.

won't do a thing for you unless you're using raw database files - with no file system associated, just the device name used as datafile name in Oracle.

> Right now we are using 4 dbwr's. I appreciate your suggestions.

Keep using the 4 dbwrs and lose the disk_async_io parameter: it's doing nothing unless you're using raw IO.

> Our UNIX SA says " Cooked Filesystems have their own Asynchronous
> Mechanisim that equally replaces Oracle Async IO"

I'll add my voice to Kevin's reply...

> What Am I missing here

A lot. To get aio in filesystems in HPUX you must use a filesystem type that supports it. As of last time I checked it, the only one that did in HP environments was vxfs, the Veritas file system for HP. None of the native HPUX ones did. Oh: did I mention that vxfs is an extra? As in: "mucho dinero"?

-- 
Cheers
Nuno Souto
in sunny Sydney, Australia
dbvision_at_iinet.net.au
--
http://www.freelists.org/webpage/oracle-l
Received on Thu Sep 07 2006 - 05:29:58 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US