Date: Wed, 3 Jun 2009 08:32:08 -0400
I agree that cacheing probably makes the second query run faster.
The way to test that is to recycle the database and repeat the test but with the test order reversed.
Of more interest to me is why one would doubt Oracle's statement on this. Analyze does not gather sufficient information for the CBO to choose the best plan. DBMS_STATS can, depending on the parameters.
What promped you to make this test in the first place?Received on Wed Jun 03 2009 - 07:32:08 CDT