Re: is SIZE AUTO ever wrong? or am I wrong?
From: Greg Rahn <greg_at_structureddata.org>
Date: Fri, 20 Nov 2009 10:47:06 -0800
Message-ID: <a9c093440911201047k26c3d894yfc20a09603d01c40_at_mail.gmail.com>
On Fri, Nov 20, 2009 at 9:05 AM, neil kodner <nkodner_at_gmail.com> wrote:
> I think it would be more efficient for letter_bin to be accessed by clm_id, and its not. It's getting accessed by prty_id and HJ to claims. Am I being unreasonable, or are my stats no good?
Date: Fri, 20 Nov 2009 10:47:06 -0800
Message-ID: <a9c093440911201047k26c3d894yfc20a09603d01c40_at_mail.gmail.com>
On Fri, Nov 20, 2009 at 9:05 AM, neil kodner <nkodner_at_gmail.com> wrote:
> I think it would be more efficient for letter_bin to be accessed by clm_id, and its not. It's getting accessed by prty_id and HJ to claims. Am I being unreasonable, or are my stats no good?
I dont know that I would worry so much about "if stats are good", I would worry more about "if execution plans are good" and then if they are not, troubleshoot systemically why.
Fist question: are the cardinality estimates for the 1) the tables and 2) the joins accurate?
If you think that a different execution plan is better, use hints to get that plan and compare them.
-- Regards, Greg Rahn http://structureddata.org -- http://www.freelists.org/webpage/oracle-lReceived on Fri Nov 20 2009 - 12:47:06 CST