Re: RAC PARALLEL

From: Sanjay Mishra <smishra_97_at_yahoo.com>
Date: Wed, 12 Oct 2011 11:14:55 -0700 (PDT)
Message-ID: <1318443295.70291.YahooMailNeo_at_web161201.mail.bf1.yahoo.com>

Tim
Thanks for the detailed update. Here is top events which are not indicating much on the gc events.

Top 5 Timed Foreground Events
Event
Waits
Time(s)
Avg wait (ms)
% DB time
Wait Class
db file sequential read 32,512,212 200,101 8 69.58 User I/O DB CPU   40,112   11.30  
db file scattered read 2,123,189 29,289 13 10.51 User I/O db file parallel read 161,091 4,058 25 2.01 User I/O PX Nsq: PQ load info query 20,702 3,689 189 1.23 Other

Sanjay

---

From: Tim Gorman <tim_at_evdbt.com>
To: smishra_97_at_yahoo.com; oracle-l_at_freelists.org Sent: Wednesday, October 12, 2011 1:51 PM Subject: Re: RAC PARALLEL

Sanjay,

RAC is an amplifier of waits on I/O and enqueues.  Assuming that you've gone "vanilla" in your 11g upgrade and not propagated forward any strange settings in initialization parameters from your previous environment, then whereever you might have seen waits on I/O (i.e. "db file sequential read", "db file scattered read", etc) in your non-RAC environment, you'll now see those waits along with corresponding RAC global cache (GCS) waits in addition.  Whereever you might have seen waits on enqueues (i.e. events beginning with "enq:") in your non-RAC environment, you'll now see corresponding RAC global enqueue (GES) waits in addition.  These additional GCS and GES waits would be the "price of RAC" from the standpoint of performance, and are to be expected.  If this is the situation you are seeing, then your response is the same in a RAC environment as in a non-RAC environment - tune the SQL to reduce I/O, and analyze/understand/minimize enqueue  activity.  As the waits on the basic I/O and enqueue events decrease, so will the waits on the dependent RAC-related GCS and GES events decrease.

The main point is that wait-event analysis will tell you whether you're experiencing "normal" additional waits on GCS and GES piled on top of the non-RAC waits you were accustomed to seeing in your non-RAC environment.  If you are seeing any other excessive waits on events, or any pattern to your waits other than that described above, then please consider posting information from an AWR report to the list (particularly the "Top Five Timed Events" section of the AWR report), and we'll do our best to help you diagnose it with you.

Hope this helps...

Thanks!

-Tim

-----Original Message-----
>From: Sanjay Mishra [mailto:smishra_97_at_yahoo.com]
>Sent: Wednesday, October 12, 2011 11:23 AM
>To: oracle-l_at_freelists.org
>Subject: RAC PARALLEL
>
>Hi

We have one environment on Sun Infrastructure where we are planning to move to RAC on 11g. So move is from non-RAC to 4-Node 11g R2 RAC. Database is OLTP, Partitioned Tables/Indexes and Apps is using heavily PARALLEL and PARALLEL_INDEX hints. What is the point to consider for the setup. Saw that we hit some big performance in Test setup and look like it may be due to Paralell setup. any good sugestion or best practices on RAC setup TIA Sanjay --
http://www.freelists.org/webpage/oracle-l

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

Received on Wed Oct 12 2011 - 13:14:55 CDT