CloudWatch
- Metrics — CPU, Network, Disk for EC2 by default (high-level). RAM is NOT collected by default — use CloudWatch Agent.
- CloudWatch Agent (Unified) — collect additional OS-level metrics (RAM, disk I/O, processes, netstat, swap) and logs.
- Custom Metrics — push your own via CloudWatch API.
- CloudWatch Logs — log groups → log streams → log events. Can send to S3, Kinesis, Lambda, OpenSearch. Logs encrypted by default.
- CloudWatch Logs Insights — query language for log analysis. Can query across multiple log groups.
- CloudWatch Alarms — 3 states: OK, ALARM, INSUFFICIENT_DATA. Targets: EC2 actions (stop/terminate/reboot/recover), Auto Scaling, SNS.
- Composite Alarms — combine multiple alarms with AND/OR. Reduces alarm noise.
- EventBridge (formerly CloudWatch Events) — schedule cron jobs, react to events from 90+ sources. Has default, partner, and custom event buses. Archive and replay events.
- CloudWatch Container Insights — ECS, EKS, Kubernetes, Fargate metrics and logs.
- CloudWatch Lambda Insights — Lambda performance monitoring (Lambda Layer).
- CloudWatch Contributor Insights — find top-N contributors from log data (e.g., top IPs).