Re: CPU service, FETCH calls wait event

Jay,

> the main wait event was "CPU service, FETCH calls". I couldn't find
> this phrase on Metalink or in the indices of both Kirti's book or Gaja's
> book. Cary's book had some information on CPU service but I'm still not
> entirely clear.

You won't see it because it is NOT a wait event. Rather, it is a "Timed" event. Following the equation that "Response Time = CPU Time + Wait Time", this is the CPU component and consists of all the CPU time required to perform parsing, block access (logical I/O) (holding/walking/waiting for latch), intrablock search, etc.... Since this is inside the CPU and thus not "waiting" for an external event, it is not a wait event. Oracle will record this under the 'c=' section in the trace files.

> Looking at cpu usage from top I see
> 8.0% idle, 54.6% user, 6.3% kernel, 31.2% iowait, 0.0% swap

You should look at the 'user' component of the individual process for this session to validate this. However, it does seem that your CPU usage is high (as compared to the 'kernel' time that is recorded for CPU times allocated to servicing OS calls.) The Higher iowait may be attributable to the SAN work that was going on. Since the Top output is a system level aggregate, we will not be able to determine how this activity contributes. The CPU time for your process should remain the same in any case...

If you are looking for tuning opportunities, look to reduce the logical I/O of the SQL involved, followed by parsing - these two are normally the biggest component of CPU usage in an Oracle process.

-- 
John Kanagaraj <><
DB Soft Inc
http://www.linkedin.com/in/johnkanagaraj
http://jkanagaraj.wordpress.com (Sorry - not an Oracle blog!)
** The opinions and facts contained in this message are entirely mine
and do not reflect those of my employer or customers **
--
http://www.freelists.org/webpage/oracle-l