Re: Solaris-Single node Vs Linux 3 nodes RAC

From: Tim Gorman <tim_at_evdbt.com>
Date: Mon, 05 Nov 2012 09:59:10 -0700
Message-ID: <5097F05E.3030106_at_evdbt.com>



Sudhakar,

Assuming that the PLAN_HASH_VALUE for SQL_ID = "9txtqr9tdkcbs" is the same in both cases, then is there any chance you could re-run the same test on one of the PDB-RAC nodes with CLUSTER_DATABASE=FALSE?

If the execution plans are different on the two environments, then you have your likely answer already, but if the execution plans are the same, then it might be useful to try to separate out any differences due to RAC versus any differences due to OS/HW.

Hope this helps...

-Tim

On 11/5/2012 8:25 AM, Sudhakar G Pillai wrote:
> All,
> I appreciate if you could shed some light on which direction to go. I have Solaris non-rac database running with 26GB SGA ( PDB-NONRAC(Solaris)) and
> 3 node RAC database in Linux with 100GB SGA ( PDB-RAC ) . One of our program runs 3 times slower on RAC vs Soalris. Same SQL code , same plan, and same oracle version etc. Solaris has file system with Veritas LVM and Linux has ASM. Here is the output of 10046:
> SQL ID: 9txtqr9tdkcbs
> Event waited on Times Max. Wait Total Waited
> ---------------------------------------- Waited ---------- ------------
> PDB-NONRAC(Solaris) db file sequential read 413468 0.32 234.99
> PDB-RAC(LINUX) db file sequential read 434173 0.50 2138.24
>
> Is it Veritas's LVM caching playing the role here vs no-caching in ASM?
>
> Thank you,
> Sudhakar G Pillai
> --
> http://www.freelists.org/webpage/oracle-l

--
http://www.freelists.org/webpage/oracle-l
Received on Mon Nov 05 2012 - 17:59:10 CET

Original text of this message