How to effectively demonstrate the system load is an essential skill when doing system tuning. Often we use the time Model provided by Oracle, for example, we need to make a trend graph similar to the following to show the system load.
Such a trend chart can be directly used by the oracle10g after the OEM, or you can pass the SQL results into Excel to make a trend map, here is not to say how to make such a diagram, but to say in our selected performance indicators, what does DB time mean? What does the DB CPU mean?
In fact, the official documentation has been explained (I hope I've noticed that): V$sess_time_model
The Event Model tree chart is well worth referencing.
In general (if there are any errors, please correct them):
1. The total time spent on the database includes background elapsed times + db timing, basically in a normal system DB time is much larger than background elapsed times (which refers to the amount that the database daemon consumes, such as the Pmon process itself).
2. DB time contains DB CPU + SQL Execute elapsed time + Parse time elapsed + other those elapsed, basically a normal system in which the first three items occupy more than 99% db SQL execute elapsed time should be more than 95%, but it's worth noting that the DB CPU and SQL Execute elapsed time are overlapping, so you'll see DB CPU + SQL Execute ELA in an AWR report psed time exceeds 100% DB time.
3. DB time is the elapsed amount (elapsed times), in subtle (microseconds) units, which is one out of 10,000 seconds. The Stat_name in V$sys_time_model is "DB time".
4. The DB CPU is the time the CPU is running, not including the time that the database process waits for the CPU, and also in microseconds (microseconds). The Stat_name in V$sys_time_model is "DB CPU".
5. The ' CPU + Wait for CPU ' we often see in the Ash report refers to DB time, and the CPU is the DB CPU.
Ext.: http://www.dbform.com/html/2010/1309.html
DB time VS. DB CPU