Re: Source for tim=nnnnn in 10046 trace

From: Gints Plivna <gints.plivna_at_gmail.com>
Date: Thu, 22 Aug 2013 16:12:43 +0300
Message-ID: <CAN2wOq1wVMD2tt_rR4Tafu6xCkiPH5-9m-wD=eCaX1B5tPwBrA_at_mail.gmail.com>



Optimizing Oracle Performance
 Autori: Cary Millsap,Jeff Holt
Chapter 5 Interpreting Extended SQL Trace Data tim
If a tim value is 0, then TIMED_STATISTICS for the session was false when the database
call time would have been calculated. You can thus confirm whether TIMED_ STATISTICS was true by observing tim values. In our field work, my colleagues
and I have found that specific non-zero tim values associated with PARSING IN
CURSOR sections are largely irrelevant.
In Oracle9i, tim is a value expressed in microseconds (1 ìs = 0.000 001 seconds).
On some systems (such as our Linux research servers), tim field values are unadulterated gettimeofday values. On other systems (like our Microsoft Windows
research machines), the origin of tim field values can be much more mysterious.
In releases prior to Oracle9i, tim is a V$TIMER.HSECS value expressed in centiseconds (1 cs = 0.01 seconds).

You can read more in
http://books.google.lv/books?id=l9eLIh2ylekC&lpg=PP1&dqÊry%20millsap%20optimizing%20oracle%20performance%20tim&hl=lv&pg=PT107#v=onepage&qÊry%20millsap%20optimizing%20oracle%20performance%20tim&fúlse

Gints

2013/8/22 rjamya <rjamya_at_gmail.com>

> Anyone know how/from where this value is sourced? is it from v$timer? it is
> time in microseconds since when?
> Plus i suspect it is platform dependent ...
> Thanks for your answers in advance
> Raj
>
>
> --
> http://www.freelists.org/webpage/oracle-l
>
>
>

--
http://www.freelists.org/webpage/oracle-l
Received on Thu Aug 22 2013 - 15:12:43 CEST

Original text of this message