Re: question on dbazine article

From: <Jared.Still_at_radisys.com>
Date: Thu, 26 Feb 2004 10:52:40 -0800
Message-ID: <OFD6477801.4FA47979-ON88256E46.006788A7-88256E46.00679EB0@radisys.com>

Re the freelists: there isn't enough information provided in the article to do
anything but guess as to why it worked.

My guess is that the add'l freelists on the index gave some relief due to a monotonically incrementing key.

Jared

"zhu chao" <chao_ping_at_vip.163.com>
Sent by: oracle-l-bounce_at_freelists.org
02/26/2004 05:16 AM
Please respond to oracle-l

        To:     <oracle-l_at_freelists.org>
        cc:     "Yong Huang" <yong321_at_yahoo.com>
        Subject:        question on dbazine article

http://www.dbazine.com/burleson20.shtml
I often visit dbazine and read articlles there, on this issue, I have some questions:
question to that article:
1. he said:
alter system set optimizer_index_cost_adj=20; alter system set optimizer_index_caching=65.

but in fact, these parameters cannot be modified online. How did he do that?

2.Implement cursor_sharing=force
According to wait event based tuning, tuning something that is not the bottleneck does not helps much. In his case, euqueue wait and full table scan caused most of the problem. Would change cursor_sharing be the solution of his problem?

3. question about add freelists;

He has 450 users inserting records, even if one person can insert a record every 3 seconds, it is only possible that there is 150 new records per second. Can't oracle process 150 record insert per second even if only 1 freelists? I did a small test with 300 concurrent session doing insert into a table, each insert a table after 3 second sleep. and this is the statspack report:( i removed the plsql locker timer event from statspack via modifying stats$idle_event).
Top 5 Wait Events

~~~~~~~~~~~~~~~~~                                             Wait     % 
Total
Event                                               Waits  Time (cs)   Wt

Time
-------------------------------------------- ------------ ------------

log file parallel write                            25,955        2,345 
90.72
control file parallel write                           146          109 
4.22
db file parallel write                                168           55 
2.13
buffer busy waits                                  30,761           34 
1.32 --only a few percent of that.
log file switch completion                              4           22 .85

Regards
Zhu chao.

Please see the official ORACLE-L FAQ: http://www.orafaq.com

To unsubscribe send email to: oracle-l-request_at_freelists.org put 'unsubscribe' in the subject line.

--
Archives are at http://www.freelists.org/archives/oracle-l/
FAQ is at http://www.freelists.org/help/fom-serve/cache/1.html


-----------------------------------------------------------------

Received on Thu Feb 26 2004 - 12:49:49 CST