Given an observability need (for example following a single request across multiple services), identify whether logs, metrics, or distributed tracing is the right approach.