Re: hash accessing with 1 LIO consumes nearly the same time as index accessing with 5 LIO.
From: <richa03_at_gmail.com>
Date: Sun, 24 Aug 2008 06:52:49 -0700
Message-ID: <26279fb90808240652t42a71e58nda005fd2727ef5f9@mail.gmail.com>
Date: Sun, 24 Aug 2008 06:52:49 -0700
Message-ID: <26279fb90808240652t42a71e58nda005fd2727ef5f9@mail.gmail.com>
Version? Platform?
Your test is only for one record and the cache is warmed - the index blocks (root, branches and leaf) and the data block will be in cache after the first itteration...
While the differance is small for this one read, the overall difference will be larger with "random" access - 10000 LIO vs 50000 LIO. The CPU cost of a column access depends upon the position of the column in the row - if the column is first, there is no CPU cost added to the plan; if the column is second or further, there is a <position>*20 added to the CPU cost. Taking this into account, I don't think the whole block is scanned...
-- http://www.freelists.org/webpage/oracle-lReceived on Sun Aug 24 2008 - 08:52:49 CDT