RE: Tuning I/O-related wait - Metalink Note: 223117.1

  • From: Andy Rivenes <arivenes@xxxxxxxx>
  • To:, <bnsarma@xxxxxxxxx>, <oracle-l@xxxxxxxxxxxxx>
  • Date: Tue, 09 Nov 2004 13:38:53 -0800

At this level (e.g. "system") response time is a misnomer. What you have is 
elapsed time and the "capacity" to consume service time up to the total, or 
elapsed time, of the interval measured. Elapsed time at this level cannot 
be accurately "calculated" by adding "service time" and "wait time". 
Elapsed time is calculated as the difference between measurement intervals. 
Wait time measurements at the system level are flawed at best due to 
interval boundary measurements and the capacity for "infinite" waiting (I 
believe Cary coined this term first). However, Lex is otherwise correct in 
my opinion, since the "wait time" will generally skew high and since it is 
only a fraction of the "service time" it is probably not an overriding 
issue. Of course, now we're venturing from workload measurement to 
performance tuning, and for that you need to look at the "process" you're 
concerned about and not focus on system level workload measurements.

At 08:28 AM 11/9/2004, Lex de Haan wrote:
>See in line ...
>Let's do some simple math from these figures:
>'Wait Time' = 10,827 x 100% / 52,01% = 20,817 cs 'Service Time' = 358,806 cs
>'Response Time' = 358,806 + 20,817 = 379,623 cs
>CPU time                    = 94.52% = 379,623 / 358,806
>direct path read            =  2.85% = 10,827  / 358,806

Andy Rivenes
Email: arivenes@xxxxxxxx


Other related posts: