Re: Background Processes and I/O

From: Billy Verreynne <vslabs_at_onwe.co.za>
Date: 21 Dec 2003 21:26:47 -0800
Message-ID: <1a75df45.0312212126.824235@posting.google.com>

"Michael J. Moore" <NOhicamelSPAM_at_comcast.net> wrote:

> From 9i Self Test ...
>
> Which 4 background processes perform I/O? (Choose 4)
>
> * DBWn
> * ARCn
> * PMON
> * SMON
> * CKPT
> * LGWR
> -----------------
> I thought they all did, except CKPT which tells DBWn to do it, and PMON
> which only works with memory structures.
>
> Their answer was all except SMON and PMON. In reality, none of them do I/O
> because they call OS library routines to do the actual I/O. This is why I
> don't have a chance in hell of passing this test, because too many answers
> depend on how you consider the question. ;-)

I agree that this is bs.

SQL> select sid, program from v$session;

SID PROGRAM

---------- ------------------------------------------------
         1 oracle_at_tjs-prod (PMON)
         2 oracle_at_tjs-prod (DBW0)
         3 oracle_at_tjs-prod (LGWR)
         4 oracle_at_tjs-prod (CKPT)
         5 oracle_at_tjs-prod (SMON)
         6 oracle_at_tjs-prod (RECO)
         7 oracle_at_tjs-prod (ARC0)
         8 oracle_at_tjs-prod (ARC1)

SQL> select * from v$sess_io where sid = 5;

SID BLOCK_GETS CONSISTENT_GETS PHYSICAL_READS BLOCK_CHANGES ---------- ---------- --------------- -------------- ------------- CONSISTENT_CHANGES

         5      46446          148181          70781         29868
                23

70,781 physical reads. This btw is the 2nd highest i/o process currently on the prod db I pulled these stats from.

Personally I think things like OCP and MSE certifications are worth crap. It is like those arses who think that a penis enlargement means they will be better at sex.

--
Billy

Received on Sun Dec 21 2003 - 23:26:47 CST