Re: How to compare two different SQL executions plans (question is more deeper then seams from first look)?

From: <J.Velikanovs_at_alise.lv>
Date: Wed, 30 Jun 2004 00:41:48 +0300
Message-ID: <OF9B9B4BBF.399C3F62-ONC2256EC2.00766A5F-C2256EC2.0077DB52@alise.lv>

Edgar,
I don't doubt we need to care about LIO from concurrency/scalability point of view.
But for CPU utilization we can’t use LIOs statistics.

Jurijs

Edgar Chupit <edgars.cupits_at_rs.lv>
29.06.2004 23:55
Please respond to Edgar Chupit  

        To:     J.Velikanovs_at_alise.lv
        cc:     oracle-l_at_freelists.org
        Subject:        Re: How to compare two different SQL executions 

plans (question is more deeper then seams from first look)?

Hello Jurijs,

JVal> Fastest from two SQL (or SQL execution plans) will be SQL with lower LIO
JVal> indicator.

We still have to remember about "scalability" (latches/locks), correct me if I'm wrong, but you still have to chose statement with lowest lio, because even if the cpu_usage is the same as with statement with larger lio count, the statement with smaller lio count will scale better. You can use package similar to T.Kytes run_stats or directly query v$mystat to see other meaningful characteristics.

I guess rule to minimize lio count can still apply, however as we all know, there is no golden rules that apply to all possible cases.

--

Best regards,

 Edgar                            mailto:edgars.cupits_at_rs.lv





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

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 Tue Jun 29 2004 - 16:46:17 CDT