Question

A CloudOps engineer manages a three-tier web application. During deployments, individual CloudWatch alarms for high CPU, elevated latency, and increased 5XX errors all fire simultaneously, generating a flood of duplicate notifications to the on-call team. The engineer wants to notify the team only when multiple related conditions indicate a genuine service degradation, while reducing alarm noise. What is the MOST appropriate solution?

Accepted Answer

Create a composite alarm that combines the existing metric alarms using an alarm rule (e.g., AND/OR logic) and send notifications only from the composite alarm

Answer

Increase the evaluation periods and datapoints-to-alarm on each individual metric alarm so they trigger less frequently

Answer

Configure a single metric math alarm that averages CPU, latency, and 5XX error rates into one aggregated metric

Answer

Add all three alarms to a CloudWatch dashboard and disable their SNS notification actions during deployment windows