Monitoring Tree

Use

The monitoring infrastructure builds a tree of monitors (a monitoring tree). Each monitor presents an agent which communicates with a particular monitored resource and processes the collected data. Typically, a monitor represents a small amount of data, that is, it contains a simple type value that gives information about a single aspect of a monitored object. For example, a monitor can represent the name of an object or the number of successful transactions with an object, but not the object itself. That is why a resource that works with several objects (for example a J2EE application which looks after the proper functioning of several banks – the resource that provides the monitoring data is the application and the logical objects that can be monitored are the different banks) and wants to provide information about them in the monitoring infrastructure has to define several monitors representing different aspects of each object (for example, the name of the bank, the number of transactions over a particular amount, and so on) and must then group them semantically so that the group represents the whole object.

A monitor is always part of a certain group of monitors describing an object. The objects themselves can be logically grouped further inside the monitoring tree in summaries; the summaries in their turn can be grouped together with other summaries; objects in summaries again, and so on. Thus, the hierarchical structure of the tree follows the semantics of the objects being monitored. In the monitoring tree, the leaves are always monitors, the nodes in the first level above the leaves are objects and the nodes in the upper levels are summaries. The summaries from the first level below the root are predefined. They are named “Kernel”, “Services”, “Applications”, “System” and “Performance” and denote different problematic areas that can be monitored inside the J2EE Engine. Typically, an application developer has to place her/his monitors and objects under “Applications”, since the others are reserved for data coming from the kernel of the server, for data describing configuration of the system (for instance, system properties) and for data coming from the services monitoring the performance of the system as a whole respectively.

The monitoring infrastructure supports 11 types of monitors, namely text, state, table, version, configuration, availability, integer, long, frequency, quality and duration monitors. For more information about each monitor type, see Types of Monitors and Their Usage.

The graphic below gives an idea what the monitor tree looks like: