Amazon DevOps Guru
ML-powered operational anomaly detection
Uses ML to analyze metrics, logs, and events from CloudWatch, CloudTrail, and Config to detect anomalous application behavior, grouping them into insights and suggesting root causes and fixes.
- Cost$$ Medium
- Difficulty●●●●●
- Costs money when idle?No
- Serverless
- In examsSOA
Think of it asLike a doctor reviewing your routine checkup results and warning you early, before you notice symptoms.
When to use it
✓ Use it when
You want to detect operational issues early without setting every alert threshold yourself.
✗ Avoid it when
You only need simple fixed-threshold alarms → CloudWatch Alarms.
Common uses
- Alerts on abnormal API latency spikes
- Detect slow RDS queries
- Suggest root causes after incidents
What does it cost?
- Pricing model: Charged by the number of resources analyzed per hour and the number of DevOps Guru API calls.
⚠️ Cost traps
Enabling it for an entire account with many resources can cost more than expected; limit it by stack or tag.
