CA APM and CA Spectrum
Beginning with version 9.1 of CA Application Performance Management (CA APM), the name "Wily" was dropped from the CA Wily Introscope® product. This section retains the name where it helps to clarify compatibility with previous versions of CA APM.
The integration of and CA APM supports versions of CA APM Introscope up to
CA APM version 9.5. However, our testing with version 9.5 did not include any APM product features that were introduced in that version.
How the Integration Works
CA Application Performance Management (CA APM) consists of an Enterprise Manager and one or more CA Introscope agents. Introscope agents are installed on servers running Java virtual machines (JVMs). The agents deliver information about numerous metrics, such as servlet response time and report metrics, to the Enterprise Manager.
Enterprise Manager uses a Management Module to organize and report on the metric data provided by an agent. By configuring thresholds on reported metrics, Enterprise Manager creates an alert when a threshold is violated and clears the alert when the threshold is no longer violated. Introscope agents are polled every 60 seconds to determine whether a threshold has been violated or whether a violated threshold has been cleared. An alert is created when a threshold event occurs.
models the CA Introscope infrastructure using two model types, the IntroscopeAdmin model and the WilyAgent model.
- The IntroscopeAdmin model represents one Enterprise Manager.
- The WilyAgent model represents an application container (JVM or CLR). When you initiate a Discovery, or when a Discovery is initiated when you restart Tomcat, requests a list of application containers from the Enterprise Manager and models them as WilyAgent models.
When CA Introscope generates an alert, data about the metric and its threshold is forwarded to . An alarm is generated on a WilyAgent model. If the alert is cleared, is notified and clears the associated alarm.
The integration with CA APM also actively monitors the inventory of application containers on an Enterprise Manager.
- When an Enterprise Manager starts to monitor a new application container, is notified and automatically creates a WilyAgent model to represent the application container.
- When an Enterprise Manager no longer monitors a discovered application container, generates an alarm on the corresponding WilyAgent model.
If you move an Introscope agent from one Enterprise Manager to another, a new IntroscopeAgent model is created and associated with the IntroscopeAdmin model. That model represents the Enterprise Manager to which the agent was moved. This activity results in two identical IntroscopeAgent models associated with two distinct IntroscopeAdmin models. Destroy the original IntroscopeAgent model so that the new IntroscopeAgent model receives all subsequent events.
The following diagram depicts the architecture of the CA APM and integration, and identifies the direction of data transfers.
- The integration combines web services with trap notifications:
- A traditional axis 1.4 polling web service that provides inventory information, such as the Management Module, the agent, alert definitions, alerts, and Management Module/Agent pairs that help determine the model. uses only the agent information.
- A bi-directional subscription web service, introscope-wssdk-consumer, uses the Apache Muse framework, which is deployed inside$SPECROOT/ tomcat/webapps. The introscope-wssdk-consumer web service listens for and processes asynchronous updates from CA Introscope.
- Trap notifications require the installation of a trap generation plugin on the CA Introscope Enterprise Manager.
- All communication between and CA Introscope passes through the Tomcat web server for agent inventory.
- The integration can be enabled on only one server.
- In a distributedSpectroSERVERenvironment, designate a OneClick server as the integration server.
You can put WilyAgent models into maintenance mode. However, the status of a WilyAgent model is dependent on event updates that the Enterprise Manager sends through SNMP traps. Therefore, before putting WilyAgent models into maintenance mode, consider the following:
- If the WilyAgent model is in an alarm state, and the Enterprise Manager posts an alarm clear update during maintenance mode, the model remains in an alarm state when it comes out of maintenance.
- If the WilyAgent model is in a normal state, and its counterpart on the Enterprise Manager enters an alarm state during maintenance mode, the model remains in a normal state when it comes out of maintenance.