Re: Performance with SCAN listener in 11g RAC
From: Guillermo Alan Bort <cicciuxdba_at_gmail.com>
Date: Thu, 2 May 2013 11:20:15 -0300
Message-ID: <CAJ2dSGTRv7jtg-6U5xbO8To=23eB1_6RDkR8hZLfWohHWAbsKQ_at_mail.gmail.com>
Did you run a trace on all the sessions? are you sure the query is exactly the same and that the generated plan is the same? are you clearing the buffer cache between runs? this sounds like the second result is cached and therefore faster...
and to answer your question, no, there is no guideline for stats gathering in RAC vs single instance, the differences are usually somewhere in the application logic (stuff that manages concurrency) or the data model design.
Date: Thu, 2 May 2013 11:20:15 -0300
Message-ID: <CAJ2dSGTRv7jtg-6U5xbO8To=23eB1_6RDkR8hZLfWohHWAbsKQ_at_mail.gmail.com>
Did you run a trace on all the sessions? are you sure the query is exactly the same and that the generated plan is the same? are you clearing the buffer cache between runs? this sounds like the second result is cached and therefore faster...
and to answer your question, no, there is no guideline for stats gathering in RAC vs single instance, the differences are usually somewhere in the application logic (stuff that manages concurrency) or the data model design.
hth
Alan.-
On Thu, May 2, 2013 at 11:10 AM, Lu Jiang <lu.jiang69_at_yahoo.com> wrote:
> Hi DBA gurus,
>
> I am do a performance test. Find that running same query using tns
> connection string with SCAN host name is 5 seconds slow than using tns
> connection string points to a specific host name (17seconds VS. 12
> seconds). Is this normal?
> Is there are any special requirement to run gather schema stats tasks in a
> RAC environment?
> Could anyone shed light on these?
>
> Thanks,
> Lu
> --
> http://www.freelists.org/webpage/oracle-l
>
>
>
-- http://www.freelists.org/webpage/oracle-lReceived on Thu May 02 2013 - 16:20:15 CEST