Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: Index Range Scans....

Re: Index Range Scans....

From: Mark D Powell <Mark.Powell_at_eds.com>
Date: 3 Mar 2006 07:30:00 -0800
Message-ID: <1141399800.707626.174980@i40g2000cwc.googlegroups.com>


You are right. But then the problem of where two columns are defined as different types but do contain the same data is another problem that I have encountered more frequently than I should have. Ususally because of implicit conversion the query runs poorly and someone asks why isn't the index being used, or some of the data in the character version of the column is not a valid number, which is what the other column definition is, and a numeric or value error results when the query is ran.

Hash joins can be disabled at the session level: alter session set hash_join_enabled = FALSE;

But usually if you code the FROM clause in the correct driving table order and then code the /*+ ORDERED */ hint you can eliminate the use of the hash join. I have found the USE_NL hint is also as often as not also required with the ORDERED hint.

HTH -- Mark D Powell -- Received on Fri Mar 03 2006 - 09:30:00 CST

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US