Effective Monitoring and Alerting

Chapter 5. Monitoring in System Automation

System metrics are a source of valuable information about a system’s state. Alarms and monitors evaluating to a Boolean value indicate the current state and operational context of your systems. Thus, you can take advantage of them to programmatically drive resilience and recoverability, while reducing the complexity of human interaction.

Self-Regulating Systems and Autonomic Computing

The holistic approach to the development of self-managing and self-regulating systems is termed Autonomic Computing (AC).

By design, an AC system does not require an operator’s attention. Instead, it implements a manager application in charge of configuration, governance, protection, optimization, and maintenance of components. The manager is meant to operate in a closed loop of four continuous steps: monitor, analyze, plan, execute. The loop opens with monitoring, reading in the information from active sensors. Gathered data serves as a base for analysis, from which the relevant action plan originates. Finally, the stabilizing action is executed and the loop starts over. That keeps the system sustainable.

The autonomic manager requires a good deal of built-in knowledge to work: the system must know what to look for, what decisions to draw from the observations, what alternatives exist for taking an action, and how to execute it. While AC is a broad subject, way beyond the scope of this book, it is well worth mentioning here, as monitoring and alerting are one ...