Re: Taking export from SECONDARY_INSTANCE from 02 node RAC
Date: Thu, 26 Feb 2009 23:51:22 +0100
Message-ID: <qr6eq453ep91042c1fi75t9eurvmhifikm_at_4ax.com>
On Thu, 26 Feb 2009 11:45:20 -0800 (PST), shweta.kaparwan_at_googlemail.com wrote:
>Actually due to application restriction we are not able to explot the
>real RAC. So we are using sort of Active-Passive mode of RAC.
>
> also could you please eloborate / guide me to some doc/ref for below
>point:
>> Importantly also a scattered read on 9iRAC for n blocks will result in n requests to the other node.
>
>Regards
>
>Shweta
I disagree. You are NOT using a 'sort of active passive mode of RAC'. In that case you would have installed Oracle RAC Clusterguard I. Obviously, you didn't.
Now you don't have any benefit from using RAC, only disadvantages. You
are flushing your money in the toilet. Please shut the second instance
down, and send me your money directly.
You should have set up Dataguard instead of RAC.
To confirm my observation you only need to do the following
alter session set events='10046 trace name context forever, level 12';
select /*+FULL */ * from <any table>;
You will observe
<db_file_multiblock_read_count> global cache cr requests after each
scattered read
in your trace file.
This feature has been removed in 10g.
-- Sybrand Bakker Senior Oracle DBAReceived on Thu Feb 26 2009 - 16:51:22 CST