RE: PARALLEL_EXECUTION_MESSAGE_SIZE

I found this page which was very interesting: http://doug.burns.tripod.com/px8.html

"Partial Message Event

• PX Deq: Msg Fragment

This event indicates that parallel_execution_message_size may be too small. Maybe the rows that are being passed between the processes are particularly long and the messages are being broken up into multiple fragments. It's worth experimenting with message size increases to reduce or eliminate the impact of this."

In this case is the presence of "PX Deq: Msg Fragment" indicates something is "broke" so there's a need to fix it. Otherwise leave it be:

>From one of my 8i databases:

SQL> r
  1* select * from v$system_event where event = 'PX Deq: Msg Fragment'

EVENT                TOTAL_WAITS TOTAL_TIMEOUTS TIME_WAITED AVERAGE_WAIT
-------------------- ----------- -------------- ----------- ------------
PX Deq: Msg Fragment         980              0          67   .068367347

So for an instance that's been up for nearly 82 days and users have waited 0.67 seconds over that whole time even though the message size is the default of ~2K. It ain't broke, I ain't fixin' it.

S-

On Wed, 5 Oct 2005, Gogala, Mladen wrote:

> No, indeed I have not tried it. I usually do not change parameters just
> because it's there.
>
> There is usually something I want to achieve by changing parameters. My main
> motto is:
>
> If it ain't broken, don't fix it. So far, I haven't seen anything that would
> benefit from changing that
>
> parameter. Besides, my database is 90% OLTP. I don't really want users to do
> parallel anything.
>
> Only DBA personnel is in consumer group which allows parallel execution. For
> everybody else,
>
> it's explicitly prohibited (max degree of parallelism=1).
>

--
http://www.freelists.org/webpage/oracle-l