Monitoring
Know when something breaks before users do: metrics, logs, traces and alerts.
About Monitoring
Monitoring and observability tell you whether a system is healthy and, when it is not, why. They rest on three kinds of signal: metrics (numbers over time), logs (records of events) and traces (one request's path through many services).
Tools like Prometheus, Grafana and the ELK stack collect and show these signals, and alerts wake someone up when something breaks.
Why it matters for placements
You cannot run production without monitoring. Interviewers ask what you would measure, how to set useful alerts, and how to find the cause of a slowdown.
What you'll learn
4 topics · 10 lessons, each with practice
- Choose the right metrics using the golden signals
- Query metrics with Prometheus and build Grafana dashboards
- Structure and search logs centrally
- Use distributed tracing to find slow services
- Write alerts that are actionable, not noisy
Foundation Foundation
Take Foundation testRequired. Enough to clear online assessments and service-company interviews.
Observability Basics
Not startedOnce code runs on servers you cannot attach a debugger, so the only way to know what is wrong is the data the system emits. This topic covers the three kinds of signals and how to set a clear reliability target. Every other monitoring tool builds on these ideas.
Interview Ready Interview Ready
Take Interview Ready testRecommended. Enough for technical rounds at good product companies.
Metrics
Not startedMetrics are cheap numbers over time, like request rate, error rate and latency, and they are the first place you look when something feels off. This topic covers collecting them with Prometheus and showing them in Grafana. Most DevOps and backend roles expect you to read and write a basic query.
Logging
Not startedMetrics tell you that something is wrong, but logs tell you what exactly happened to one request. This topic covers writing logs that machines can search and collecting them from many servers into one place. Without this, debugging a service with many containers means guessing.
Advanced Advanced
Take Advanced testOptional. For top product companies. Never counted against your readiness.
Alerting
Not startedNobody watches dashboards all day, so alerts are how a team finds out that users are hurting. This topic covers writing alerts worth waking up for, running an on-call rotation and using error budgets to balance speed against stability. Bad alerting either misses outages or buries people in noise.