Real-time Connection Monitoring Best Practices › Long Running Connections

Long Running Connections

Identify any applications that rely on one or more long running connections, and define an event detector to identify issues associated with these connections.

Business Value:

Critical business applications sometimes rely on specific IP connections being always in existence. The loss of one or more of these connections can result in serious impact to the business application and the service it provides. The quick identification of a long running IP connection ending can allow for automated or operator assisted recovery to minimize the business impact.

Additional Considerations:

The appropriate event detector to use depends on the monitoring requirements, for example:

• When a specific number of connections must always be present, use a CONNSTAT event detector, which enables you to use the minimum number of active connections as a criterion. This detector verifies that the minimum number of connections are present at startup and highlights any subsequent failures that lead to the minimum number not being available.
• When there are not a specific number of connections required but any started connections must remain active, use a TCPEND event detector, which supports the detection of any connection ending. If you are only interested in abnormal terminations, you can specify a termination reason code.

  For TCPEND event detectors, we recommend that you specify alerting by server to consolidate all failures for an application under a single alert.

More Information:

For information about how to define event detectors, see the Implementation Guide.