Re: Logical IO

  • From: Yong Huang <yong321@xxxxxxxxx>
  • To: john.o.golden@xxxxxxxxx
  • Date: Thu, 19 Nov 2009 08:22:48 -0800 (PST)

> I ever tried to trace such CR events as 10200, but found
> they did not log the reads on some special blocks, such 
> as segment header. 

You're right John. A recent discussion on another forum concluded 
that the only way to find the current mode read is either use event 
10046 (when the blocks are not already cached) or check x$bh for 
increase in touch count (class 4 for segment header, verify with 
file and block number). There's no event to trace CU read.

In addition, it appears that since 9iR2, event 10046 or SQL trace 
counts current mode read into CR read statistics for a simple query. 
You can do a test or see "Followup   April 26, 2005 - 8am Central 
time zone" of

http://asktom.oracle.com/pls/asktom/f?p=100:11:0::::P11_QUESTION_ID:865586003021

Only select for update or DML shows "db block gets" in sqlplus 
autotrace (or equivalent in 10046 trace summary). See the last 
paragraph of

http://yong321.freeshell.org/oranotes/SegmentHeaderCR.txt

Yong Huang


      
--
//www.freelists.org/webpage/oracle-l


Other related posts: