[ SRE_ARENA ]
← BACK TO ROADMAP
OBSERVABILITY (CLOUDWATCH)

Observability: CloudWatch, CloudTrail & X-Ray

Visualization: 1. The Three Pillars of Observability
DOWNLOADING CONCEPT: 1. THE THREE PILLARS OF OBSERVABILITY...
  • ↳You cannot manage what you cannot measure.
  • ↳AWS provides a native suite of tools to observe your infrastructure:
  • ■CloudWatch Logs: You should never SSH into a server to read a log file. The CloudWatch Agent streams your local app.log files into a centralized, searchable database.
  • ■CloudWatch Metrics: AWS automatically tracks basic data (CPU, Network In/Out). You can also send *Custom Metrics* (e.g., 'FailedLogins').
  • ■CloudWatch Alarms: The brain of the operation. Alarms constantly evaluate Metrics. If CPU stays above 80% for 5 minutes, the Alarm fires and tells the Auto Scaling Group to boot another server.
▲ LIVE INCIDENTA junior developer is confused because they deployed a brand new Docker container to ECS, but they don't see any RAM (Memory) Utilization metrics in CloudWatch, only CPU metrics.
sre@arena:~/cloud-obs
sre@arena:~$ aws cloudwatch list-metrics --namespace AWS/EC2