eG Help

Oracle's response time for an operation is composed of time executing (=CPU time) and time spent waiting (=Waiting time). An increase in either or both the above-mentioned factors will adversely impact the responsiveness of the Oracle cluster service. When Oracle executes an SQL statement, it is not constantly executing. Sometimes it has to wait for a specific event to happen before it can proceed. For example, if Oracle (or the SQL statement) wants to modify data, and the corresponding database block is not currently in the SGA, Oracle waits for this block to be available for modification. The Waiting time refers to the time spent by the Oracle server waiting for such events to complete. Oracle has a bunch of events that it can wait for - eg., buffer busy waits, db file scattered read, db file sequential read.

Whenever users complaint of a slowdown while accessing databases in a cluster, it would be helpful to know which node is experiencing a slowdown and where it is spending too much time - is the time executing more than the time spent waiting, or vice-versa? To determine this, you should monitor both the CPU time and the Waiting time of each node of the cluster. This test enables you to perform ‘half’ this analysis. In other words, this test reports the percentage of time that every node spent on waiting for one/more events to complete. This way, the test helps you understand whether/not the waiting time is contributing to the poor responsiveness of the cluster service and which node has been waiting too long.

Outputs of the test : One set of results for each node in the Oracle cluster being monitored.

Measurement	Description	Measurement Unit	Interpretation
DB_timewait_prcnt	Indicates the percentage of time this node spent on waiting for one/more events to complete.	Percent	A high value is indicative of the following cases: An increase in load (either more users, more calls, or larger transactions) I/O performance degradation (I/O time increases and wait time increases, so DB time increases) Application performance degradation CPU-bound host (foregrounds accumulate active run-queue time, wait event times are artificially inflated)