Amazon CloudWatch Application Signals
APM monitoring for applications
CloudWatch's APM capability: automatically collects metrics and traces (based on OpenTelemetry) for applications, shows latency, errors, and traffic per service, and lets you set SLOs.
- Cost$$ Medium
- Difficulty●●●●●
- Costs money when idle?No
- Serverless
- In examsDVA · SOA
Think of it asLike a hospital board showing each department's status: you instantly see which one is overloaded.
When to use it
✓ Use it when
You need to track microservice health and SLOs without building your own APM system.
✗ Avoid it when
Only need basic infrastructure metrics → regular CloudWatch metrics.
Common uses
- Set a latency SLO for a payments API
- Pinpoint which service causes errors in a microservice chain
- Monitor apps on EKS, ECS, EC2, Lambda
What does it cost?
- Pricing model: Charged by the volume of application signals (requests tracked) and traces collected.
⚠️ Cost traps
High-traffic apps generate many signals and traces, and costs grow accordingly — consider sampling.
