SYSTEM EVENTS
A crude check to determine how the Oracle instance is behaving overall can be found in the v$system_event view as shown below.
SELECT event, total_waits, total_timeouts, time_waited, average_wait
FROM v$system_event
The statistics in this view are cumulative and are representative from the time the instance was started. As a result, some of the statistics may be averaged out over time and the problem may not appear or be as obvious. In addition, many of these events are idle wait events or occur infrequently and are not important. The more important wait events are:
1. latch free
2. buffer busy waits
3. db_file scattered read
4. db file sequential read
5. library cache pin
The total_waits columns represents the total number of times a process has waited for this event since the database was started.
The total_timeouts represents the total number of timeouts a process encountered waiting from the event.
The time_waited and average_waited columns show the cumulative and average times respectively spent by the process waiting for this event in centi-seconds.
Unfortunately there are no simple numbers that can be thrown out to help in the identification of a problem area. Each instance has its own characteristics and patterns. However, once these patterns are understood, they can be used for future identification of problems if properly benchmarked.
Comments are currently closed.