CloudWatch alarms aggregate metrics over a period of time and automatically initiate actions (typically, notifications). The aggregate value needs to consistently breach a predefined threshold and be sustained over a period of time for an alarm to be triggered. This filters out the false positives and temporary glitches in the metric values and ensures that the alarms are strictly actionable.
Some examples of alarms are as follows:
- One can configure an alarm to activate when the memory utilization of a computer resource stays consistently over a safe value (say, 75%)
- One can create billing alarms for when the charges for resource utilization go over a particular value (the budget set for IT)
