Closely associated with the time model section of the report are the wait class and wait event statistics sections.

Within Oracle, the duration of a large number of operations (e.g. Writing to disk or to the control file) is metered. These are known as wait events, because each of these operations requires the system to wait for the event to complete. Thus, the execution of some database operation (e.g. a SQL query) will have a number of wait events associated with it. We can try to determine which wait events are causing us problems by looking at the wait classes and the wait event reports generated from AWR.

Wait classes define "buckets" that allow for summation of various wait times. Each wait event is assigned to one of these buckets (for example System I/O or User I/O). These buckets allow one to quickly determine which subsystem is likely suspect in performance problems (e.g. the network, or the cluster).

In this report the DB CPU wait class has the largest number of waits (total of 1961 seconds)

Wait events are normal occurrences, but if a particular sub-system is having a problem performing (e.g. the disk sub-system) this fact will appear in the form of one or more wait events with an excessive duration.

The wait event report then provides some insight into the detailed wait events. Note that this section is sorted by wait time (listed in microseconds).