User Tools
Site Tools
Know your environment - Get the performance fingerprint of your database application
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
Key performance indicators
Typical performance indicators:
- All the system statistic Values
- Logon rate
- Active Session Count
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.
Example
See this small example for the Oracle statistic value “Physical write total bytes”.
With only the pure values the information is not very useful:
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:
Collect the Data
One tool you can use to collect all this data is Splunk, Splunk is a featured, platform for collecting, searching, monitoring and analyzing machine data.
To collect the data from the database I use the plugin (app) Splunk DB Connect .
