KCNA - Cloud Native Architecture - Section 4.2

Explain observability for cloud native systems and the signals it is built on.

Metrics, logs and traces as the three primary signals; Prometheus and its pull-based scrape model; OpenTelemetry as the vendor-neutral instrumentation standard; Fluentd for log collection; and the difference between monitoring and observability. Observability sits inside this domain on the current outline.

metrics, logs and tracesPrometheusOpenTelemetryFluentdJaegerSLI, SLO and SLA

Practice question for this objective

Free sampleCloud Native Architecturemedium

Observability for cloud native systems is commonly described as resting on three primary signal types. A team wants to record the exact ordered path a single user request took as it hopped across several microservices, including how long it spent in each service. Which of the three signals is designed to capture that per-request cross-service journey?

  • AMetrics, because they aggregate numeric measurements sampled over time into series that can be graphed and alerted on.
  • BLogs, because each service writes a timestamped text record that can be searched after the fact.
  • CTraces, because they follow one request end to end and record a span for each service it passes through with per-service timing. Correct
  • DEvents, because they publish a discrete notification whenever a service changes state during the request.
Distinguish the three core observability signals and recognise that traces capture a single request's ordered path across services. A distributed trace models one request as a tree of spans, one per service or operation, each carrying start time and duration, so the request's ordered route and per-hop latency can be reconstructed. Metrics aggregate and logs are discrete records, so neither reconstructs a single request's cross-service path.

Why A is wrong: Metrics are tempting because they do capture latency numbers, but they aggregate values across many requests and cannot reconstruct the ordered path of one specific request through the services.

Why B is wrong: Logs are tempting since each service does emit timestamped records, but without correlation they are discrete events per service and do not by themselves reconstruct one request's ordered cross-service path.

Why C is correct: Distributed traces are exactly the signal for this: they stitch together spans from each service a single request touches, preserving order and per-hop duration.

Why D is wrong: Events sound relevant because they mark state changes, but they are not one of the three core observability signals and do not record a request's full ordered journey with per-service timing.

See more KCNA practice questions, answers explained.

More in this domain

Back to all Cloud Native Architecture objectives, or the KCNA cert hub.

Examworthy is not affiliated with or endorsed by Cloud Native Computing Foundation. Original, blueprint-aligned practice material only.