Re: SQL ordered by Physical Reads (UnOptimized)
From: Tanel Poder <tanel_at_tanelpoder.com>
Date: Sat, 21 May 2011 17:25:55 +0300
Message-ID: <BANLkTi=jMEXUAR5+HBbxc-xsqqgqz3D5Dw_at_mail.gmail.com>
These slides fail to mention that the optimized reads statistic is also incremented when smart scans can "optimize IO" by avoiding doing it thanks to storage indexes. So, if your OLTP database has some smart scanning happening in it (usually thanks to various reports, etc) then the optimized reads may be either due to flash cache hits or storage index IO elimination.
Date: Sat, 21 May 2011 17:25:55 +0300
Message-ID: <BANLkTi=jMEXUAR5+HBbxc-xsqqgqz3D5Dw_at_mail.gmail.com>
These slides fail to mention that the optimized reads statistic is also incremented when smart scans can "optimize IO" by avoiding doing it thanks to storage indexes. So, if your OLTP database has some smart scanning happening in it (usually thanks to various reports, etc) then the optimized reads may be either due to flash cache hits or storage index IO elimination.
-- Tanel Poder http://blog.tanelpoder.com On Sat, May 21, 2011 at 6:20 AM, Karl Arao <karlarao_at_gmail.com> wrote:Received on Sat May 21 2011 - 09:25:55 CDT
> Kyle,
>
> There's a good explanation about that metric here
>
> Oracle Exadata and OLTP Applications
>
> http://www.oracle.com/webfolder/technetwork/Exadata/MAA-BestP/OLTP/021511_93778_source/index.htm
>
> on slide #15 and #16
>
>
-- http://www.freelists.org/webpage/oracle-l