Understanding timers

Timer sensors build an aggregate of all calls to a method. They provide the number of times the method was called, the average response times and min/max response times. Timers are also used within invocation sequences to user-transaction centric timings. It is important to understand the difference. Timers without being viewed within invocation sequences aggregate all calls to a method within a certain interval. Timer that are called within an invocation sequence additionally provide the concrete time this one invocation took. The invocation sequence sensor collects these single calls and builds up a hierarchy based on them.

Note that from version 1.3 all the timer data sensors are sending to the CMR, will be additionally aggregated in order to decrease the number of insertions to the database where historical data is stored. It does not matter if the timer data was within or out the invocation sequence, it will be processed by the "aggregator" and stored in intervals of 10 seconds. The data in the CMR memory buffer will stay as it was received from the agent, thus not aggregated. This means that the timer graph will display data in groups in space of 10 seconds. Also note that the historical data can be viewed even with the restart of the CMR, which is not the case with the data in memory buffer.

Using timers for monitoring

Standard timers are represented as graphs within the inspectit client. Browse the instrumentation browser to the method on which a Timer is configured and double-click it opens the graph representation on the detailed screen. Timers are especially useful for application monitoring as they impose only minimal load on the system.

Autorefresh is disabled by default

By default auto-refresh is disabled to reduce the load on the CMR. If you want to monitor a timer you can either refresh it manually or enable auto-refresh

Method aggregation

At the lower part of the view, each graph also provides additional information about the data it displays. This aggregation is useful for certain scenarios.

For example using the historical data view (more information) and setting the timeframe to one hour, the analyst can see how often the method was called, what was its average in this time span and how long did all methods added together take (duration)

Trend analysis using historical data facility

For each view within inspectIT historical data can be viewed. Arguably this feature is most important to timer monitoring. For more details on how historical data view can be activated click here