Re: AWR interpretation

From: Andy Klock <andy_at_oracledepot.com>
Date: Wed, 30 Oct 2013 15:23:29 -0400
Message-ID: <CADo_RaPstPAZOyW4NtsB6K0uvMYxHRyvPP-RigZ9s_NhuFLEVw_at_mail.gmail.com>



On Wed, Oct 30, 2013 at 2:55 PM, Storey, Robert (DCSO) <RStorey_at_dcso.nashville.org> wrote:
>
>
> Top 5 Timed Foreground Events
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> Avg
> wait % DB
> Event Waits Time(s) (ms) time Wait Class
> ------------------------------ ------------ ----------- ------ ------ ----------
> log file sync 7,176,907 5,301 1 61.8 Commit
> DB CPU 3,122 36.4
> db file sequential read 20,772 129 6 1.5 User I/O
> SQL*Net message to client 14,386,916 17 0 .2 Network
> control file sequential read 4,186 12 3 .1 System I/O
>
> Host CPU (CPUs: 4 Cores: 4 Sockets: 2)
> ~~~~~~~~ Load Average
> Begin End %User %System %WIO %Idle
> --------- --------- --------- --------- --------- ---------
> 12.3 7.3 80.4
>

That is easier to read :) You mentioned that you are committing (by array) COMMIT=Y, so what is your BUFFER set to? The log file sync's are indicative of too many commits which in this case probably aren't warranted. So crank up the BUFFER or set COMMIT=N.

Three log switches a minute are pretty frequent and will probably ramp up even more when you reduce those log file sync's.

Andy K

--
http://www.freelists.org/webpage/oracle-l
Received on Wed Oct 30 2013 - 20:23:29 CET

Original text of this message