System Design Lab on YouTube · Open on YouTube
Running production: reliability, observability, performance and security · Proficient · Lesson 70 of 98 · 36 min video, then about 8 hours of practice
Observability: logs, metrics and traces
After this you can instrument a service with structured logs, metrics and distributed traces, and use them together to find the cause of a problem.
Why it matters
Observability is being able to ask new questions of a running system without shipping new code. Logs tell you what happened, metrics tell you how much and how often, traces show you where time went across services. The power is in correlating them. Learn the golden signals (latency, traffic, errors, saturation), the difference between counters, gauges and histograms, and how OpenTelemetry gives you vendor-neutral instrumentation. Trap: high-cardinality labels on metrics (user IDs, request IDs). They will explode your monitoring bill.
How to use this resource
A complete, current crash course. Watch it all; 0:20:41–0:24:22 on SLOs and alerting connects to the next topic.
Go further
Go deeper