When production breaks at 3am, dashboards are the difference between a five-minute fix and a five-hour outage. Observability rests on three complementary pillars.
1. Logs — What Happened
Structured, searchable records of discrete events. Log JSON (not plain text), include a request/correlation ID, and ship to a central store so you can search across all instances.
2. Metrics — How Much / How Fast
Numeric time-series: request rate, error rate, latency percentiles (p95/p99), CPU, memory. Cheap to store and perfect for dashboards and alerting on trends.
3. Traces — Where the Time Went
A trace follows one request across services, showing how long each hop took. Indispensable for finding the slow database call buried in a chain of microservices.
Standardize on OpenTelemetry
OpenTelemetry is the vendor-neutral standard for all three signals. Instrument once, then send data to Grafana, Datadog, Honeycomb, or whatever you choose — no rewrites.
