The three pillars of observability are?

The three pillars of observability are?

• A. Errors, logs, and metrics
• B. Logs, events, and metrics
• C. Logs, metrics, and traces
• D. Metrics, traces, and events

Answer: (C)

Observability rests on three kinds of data that together reveal how a system behaves: logs, metrics, and traces. Metrics are the numeric, time-stamped measurements you watch for trends—like request rate, error rate, and latency percentiles—used for dashboards and alerts. Logs capture detailed, timestamped events and messages from applications and infrastructure, giving context for what happened at a specific moment. Traces show the path a single request takes across services, with timing information for each hop, so you can see how a request propagates and where delays occur. This combination is why logs, metrics, and traces are the best fit: metrics provide quick health signals, logs give detailed incident context, and traces illuminate the end-to-end flow across distributed components. The other options mix in concepts (like errors or events) that don’t by themselves provide the full cross-service visibility that traces offer, or replace traces with something that doesn’t reveal request flows.

Observability rests on three kinds of data that together reveal how a system behaves: logs, metrics, and traces. Metrics are the numeric, time-stamped measurements you watch for trends—like request rate, error rate, and latency percentiles—used for dashboards and alerts. Logs capture detailed, timestamped events and messages from applications and infrastructure, giving context for what happened at a specific moment. Traces show the path a single request takes across services, with timing information for each hop, so you can see how a request propagates and where delays occur.

This combination is why logs, metrics, and traces are the best fit: metrics provide quick health signals, logs give detailed incident context, and traces illuminate the end-to-end flow across distributed components. The other options mix in concepts (like errors or events) that don’t by themselves provide the full cross-service visibility that traces offer, or replace traces with something that doesn’t reveal request flows.