CONCEPTS
Monitoring
The process of collecting, analyzing, and using data to track the health of applications and infrastructure.
Monitoring
The process of collecting, analyzing, and using data to track the health of applications and infrastructure.
"The Dashboard"
Monitoring is about knowing the knowns. It answers: "Is the system healthy according to the rules we defined?"
Blackbox vs. Whitebox
- Blackbox Monitoring: Testing from the outside. "Is the website returning 200 OK?" (Pingdom, UptimeRobot).
- Whitebox Monitoring: Testing from the inside. "Is the JVM heap usage < 80%?" (Prometheus, CloudWatch).
The Golden Signals (Google SRE)
If you monitor nothing else, monitor these four:
- Latency: Time to serve a request.
- Traffic: Demand on the system (RPS).
- Errors: Rate of failed requests.
- Saturation: How "full" the service is (CPU, Memory, Disk).
ExThe Silent HDD Failure
"A database server's hard drive filled up. The "Disk Full" alert was disabled because it was "noisy"."
Impact
The database locked up, causing a 6-hour outage.
Resolution
The team re-enabled the alert but tuned it to fire on "Projected to fill in 24 hours" rather than "95% full", giving them time to react.
Why Monitoring Matters
Monitoring is the foundation of incident detection. Without it, you rely on customers to tell you something is broken.
Good monitoring means knowing about problems before your users do.
Common Pitfalls
Vanity Metrics
Monitoring "Total Requests Served" looks nice on a slide, but "Error Rate" is what tells you if you are broken.
How to Use Monitoring
๐ฏ
Four Golden Signals: Latency, Traffic, Errors, Saturation.
๐ค
Synthetics: Simulate user journeys.
๐งน
Clean Dashboards: Delete charts that no one looks at.
Related Terms
Frequently Asked Questions
Monitoring vs. Observability?
Monitoring is for **known problems** (Is CPU high?). Observability is for **unknown problems** (Why is checkout slow only for iOS users in Germany?).