Message-Id: <10628.117725@fatcity.com>
From: Don Dealy II <dondealy@teleport.com>
Date: Sat, 23 Sep 2000 07:37:12 -0700
Subject: Performance:  table access by rowid range

I've been looking at the explain plans on a 3-rd party package that we are 
using.  I see that every single query is using "table access by rowid 
range" with 2 bind variables.  Some of these queries have additional 
"where" clauses, but those on indexed columns are always defeated because 
of a function to the left of the operator, ie upper().

In looking at various examples in the Oracle docs about "table access by 
rowid", all of the examples always have the "by rowid" working directly 
with an index.  Since this app is ignoring or defeating all of the indexes, 
is there any performance difference between a full table scan and "by rowid 
range"?