5 Replies Latest reply: Feb 7, 2013 1:10 AM by Nikolay Savvinov RSS

    Time model/event


      what is time model stats in AWR and time event in 10g?
      What is the difference between two?

        • 1. Re: Time model/event
          user13809017 wrote:

          what is time model stats in AWR and time event in 10g?
          What is the difference between two?
          Do you realize that we are not standing behind you & have no idea what you are looking at?

          How do I ask a question on the forums?
          SQL and PL/SQL FAQ
          • 2. Re: Time model/event
            not sure what you're asking, but maybe start here

            • 3. Re: Time model/event
              I guess that time model is cumulative time which is collected from all non-idle database sessions. Time event is percentage of time spend by database in actual execution.
              • 4. Re: Time model/event

                Below is a description for Time model statistics in AWR. I do not understand what you mean by time waits in AWR - Pls clarify!

                Note: Below text is a blatant copy of what Don has to sya on this in his website. This is the simplest description I have got. To look for further info, kindly look into the good old Oracle documentation.
                Time Model Statistics
                The time model statistics give insight about where the processing time is actually spent during the snapshot interval.
                Time Model Statistics  DB/Inst: LSQ/lsq  Snaps: 1355-1356
                -> ordered by Time (seconds) desc
                                                                       Time      % Total
                Statistic Name                                     (seconds)     DB Time
                ----------------------------------------- -------------- -----------
                DB time                                         7,274.60      100.00
                sql execute elapsed time                        7,249.77       99.66
                background elapsed time                           778.48       10.70
                DB CPU                                            150.62        2.07
                parse time elapsed                                 45.52         .63
                hard parse elapsed time                            44.65         .61
                PL/SQL execution elapsed time                      13.73         .19
                background cpu time                                 8.90         .12
                PL/SQL compilation elapsed time                     3.80         .05
                connection management call elapsed time              .15         .00
                Java execution elapsed time                          .05         .00
                hard parse (bind mismatch) elapsed time              .00         .00
                hard parse (sharing criteria) elapsed time           .00         .00
                sequence load elapsed time                           .00         .00
                failed parse (out of shared memory) elapsed          .00         .00
                inbound PL/SQL rpc elapsed time                      .00         .00
                failed parse elapsed time                            .00         .00
                In the sample output of the AWR Time Model Statistics Report shown above, it can be seen that the system spends the most processing time on actual SQL execution but not on parsing.
                • 5. Re: Time model/event
                  Nikolay Savvinov

                  time model stats is more of a high-level breakdown of where the database spends most of the time. For example, it can tell you how much time you spend parsing queries compared to time executing them.

                  Timed events info is more detailed and specific, it can tell you more about where the database is spending time and why. For example, it can tell you that you're spending a lot of time parsing because there is concurrency for low-level serialization structures in the shared pool (mutexes, locks, latches), or that you're spending a lot of time executing SQL because your I/O is slow, etc.

                  I find that time model stats is helpful in a limited number of scenarios (such as parsing problems, logon storms etc.) while timed events lie in the heart of all basic database performance tuning activities.

                  Best regards,