I find that the most confusion is caused by "direct read/write"
events because they're caused not only by sorts but also by
PQ processes which read directly from the file into PGA, bypassing
the buffer cache. People have been quick to establish new myths,
which tell us that "scattered I/O = full table scan, sequential
disk read = index read". Direct I/O is perplexing, if you happen to
be a believer.
--
Mladen Gogala
Oracle DBA
email:mladeng_at_aetvn.com
Ext: 9787
> -----Original Message-----
> From: Cary Millsap [mailto:cary.millsap_at_hotsos.com]
> Sent: Thursday, August 19, 2004 10:28 AM
> To: oracle-l_at_freelists.org
> Subject: RE: db_file_mutliblock_read_count and physical IO
> My concern is that people who have learned ridiculous
> nonsense like the "80/20 rule of indexing" and that "Oracle
> reads records from disk into = RAM" might also not understand
> that the blocks being read can be from = segments of
> different types. In particular, I find people to be somewhat
> surprised = on occasion when they see an undo block being
> read via one of the kernel's = read events.
----------------------------------------------------------------
Please see the official ORACLE-L FAQ: http://www.orafaq.com
----------------------------------------------------------------
To unsubscribe send email to: oracle-l-request_at_freelists.org
put 'unsubscribe' in the subject line.
--
Archives are at http://www.freelists.org/archives/oracle-l/
FAQ is at http://www.freelists.org/help/fom-serve/cache/1.html
-----------------------------------------------------------------
Received on Thu Aug 19 2004 - 09:38:35 CDT
