From Chapter 1's description of ORTA, you know that response time is the sum of service time and queue time. From the previous discussion of the wait event views, you know that through Oracle's wait interface, you can collect very detailed queue time information. Oracle also allows service time collection with some level of breakdown. We can combine our objectives to simplify and to effectively communicate response time by categorizing the detailed time components.

Figure 2-14 illustrates various levels of Oracle-focused time classification. This information could represent the results of an entire Oracle system over a one-hour period. It could also represent the results of profiling a single transaction, a single session, or perhaps a group of sessions. Regardless, the diagnostic and communication aid provided by such time classification are significant. The benefit increases when combined with a little information, such as available CPU capacity, workload activity, and a response-time graph.

Notice there are a number of classification levels, allowing you to pick the most appropriate for the given occasion. When combined with Oracle's percentage of the operating system CPU consumption, level 2 provides insights into the CPU subsystem utilization, without running a single operating system command! You'll learn more about this in Chapter 5. When talking with management, I usually work at level 3, and if the managers can grasp more detail, I'll go to level 4. When talking with DBAs and devising solutions, I'll be working at level 5.