What should the solutions architect do to meet these requirements?
Create Amazon CloudWatch composite alarms where possible.
Create Amazon CloudWatch dashboards to visualize the metrics and react to issues quickly.
Create Amazon CloudWatch Synthetics canaries to monitor the application and raise an alarm.
Create single Amazon CloudWatch metric alarms with multiple metric thresholds where possible.
Explanations:
Composite alarms allow the combination of multiple CloudWatch alarms, enabling the architect to trigger an action only when both CPU utilization exceeds 50% and high read IOPS occur, thus reducing false alarms and meeting the specific requirements.
While dashboards provide visualization and quick issue detection, they do not implement alarms or automated actions based on metric thresholds. Therefore, they do not fulfill the requirement of taking action based on CPU and IOPS metrics.
Synthetics canaries are designed for application monitoring rather than infrastructure metrics. They do not specifically address CPU utilization or IOPS thresholds, so they are not suitable for this requirement.
Single alarms with multiple metric thresholds may not effectively correlate CPU utilization and IOPS metrics. This could lead to false alarms or missed actions, as it doesn’t consider the simultaneous conditions required for action.
In my opinion, the answer is:
Create Amazon CloudWatch composite alarms where possible.