Learning Objectives
By the end of this chapter, you will be able to:
- Implement metrics, alarms, monitoring, and logging across AWS workloads, including alarm-invokable actions and EBS performance optimization.
- Apply CloudWatch, CloudTrail, and Amazon Managed Service for Prometheus to monitor serverless, compute, and AI workloads, and optimize S3 transfer performance.
- Configure and manage the CloudWatch agent across EC2, ECS, and EKS, automate remediation with Lambda and Systems Manager (including Kiro and AWS DevOps Agent), and manage shared storage on EFS, FSx, and S3 Files.
- Create CloudWatch alarms and composite alarms, route events with Amazon EventBridge (input transformers and Pipes), and tune database performance with Amazon RDS Performance Insights and Amazon RDS Proxy.
- Create CloudWatch metric filters from logs, author Systems Manager Automation runbooks, and optimize EC2 with placement groups.
- Build customizable and shareable CloudWatch dashboards across accounts and Regions, and pick the right Amazon S3 storage class for a workload.
- Configure SNS notifications from AWS services and alarms, and apply right-sizing recommendations from Compute Optimizer, Cost Explorer, Budgets, the Cost and Usage Report, and Trusted Advisor.
- Troubleshoot CloudWatch alarms (including INSUFFICIENT_DATA), and apply resource tagging plus tag-based cost allocation.
Executive Summary
- CloudWatch is the central observability service: it stores metrics, runs alarms, hosts dashboards, indexes logs (Logs Insights), and emits change events that EventBridge consumes for automation.
- Alarms invoke either a direct AWS action (EC2 stop, reboot, recover, terminate; Auto Scaling; Systems Manager OpsItem; SNS) or an Amazon EventBridge event that fans out to Lambda, Step Functions, ECS tasks, or Systems Manager Automation.
- The CloudWatch agent is the bridge for telemetry that the AWS hypervisor cannot see: memory, disk-space, custom application metrics, and structured logs from EC2, ECS, and EKS hosts.
- Cost and performance optimization is data-driven: Compute Optimizer right-sizes EC2, Lambda, EBS, and Auto Scaling groups; Cost Explorer and the Cost and Usage Report quantify spend; Trusted Advisor catches over-provisioning and idle resources; S3 lifecycle policies and storage-class selection cut storage spend without rewriting applications.
Assumptions
- Workloads run in commercial AWS Regions; GovCloud and China partition feature gaps are out of scope.
- Service quotas referenced are default account quotas before requesting an increase.
- All command examples use the AWS CLI v2 syntax; older v1 invocations may differ.
- Terminology follows the official AWS documentation linked in the sources block.