If you search for performance bottleneck or tune SQL statements the system statistic of the Oracle database comes more and more out of the focus.
But these statistics are still very helpful to get a statistical overview over the runtime behavior of your database. In a standard OLTP application these statistic shows over some time a typical load behavior.
A typical problem for the database administrator is the rumor that the database is today slower than yesterday.
Without such statistic is it very complex to prove the opposite. And may be the user is right and we have relay a problem
Typical performance indicators:
But also the typical performance values of your operation system, like CPU usage, memory, CPU queue depth.
To interpret these values you need the counters per interval.
See this small example for the Oracle statistic value “Physical write total bytes”.
Let us take a look on the read count as differences from one point in time to the other:
But if we now calculate the values over the time, we get more interesting data over the runtime behavior of the application: