What is observability (as distinct from traditional monitoring), and what are its three commonly cited pillars?

Observability is a system's property of exposing enough internal state that engineers can diagnose novel, unanticipated problems from external outputs -- the three pillars are logs (discrete events), metrics (aggregated numerical measurements), and traces (following a request's path across distributed services).

logger.info({ event: 'order_placed', orderId, durationMs }, 'Order placed');
metrics.increment('orders.placed');
const span = tracer.startSpan('process-payment');
span.end();