Re: IO testing - CALIBRATE_IO (11g) vs. other (10g)
Date: Thu, 4 Apr 2013 21:11:34 -0400
Message-Id: <19EC9DFA-E1B4-4FD0-9BA5-D2D21C911E8E_at_gmail.com>
To follow up on Kellyn's mention below, you need to run NOWORKLOAD system stats during a very low/idle time before gathering stats during a workload. There are certain statistics (in 10g at least) that aren't calibrated except during a NOWORKLOAD capture.
The noworkload takes about 5 (?) minutes to execute.
Chris
Sent from my mobile device. Please ignore any typos.
On Apr 4, 2013, at 6:15 PM, Kellyn Pot'vin <kellyn.potvin_at_ymail.com> wrote:
> FYI...Callibrate_IO is only available to auto DOP, even though you'll see the data presented in 10053 trace files as the CBO is going throigh the motions, unless auto DOP kicks in, the data is useless to Oracle. All workload IO data provided to the database, even in 11g must be done with system statistics during a solid workload period. It may make a difference or may offer no benefit at all. A change in hardware after migration would be my recommendation to update the system stats.
> Hopw this helps,
> Kellyn
> Sent from Yahoo! Mail on Android
>
>
> --
> http://www.freelists.org/webpage/oracle-l
>
>
-- http://www.freelists.org/webpage/oracle-lReceived on Fri Apr 05 2013 - 03:11:34 CEST