PCA case studies
PCA · Put your knowledge to work in context.
The Prometheus Certified Associate (PCA) is an entry-level, pre-professional certification from CNCF that validates foundational knowledge of observability and hands-on familiarity with Prometheus, the open source systems monitoring and alerting toolkit. It is aimed at engineers and application developers with a special interest in observability and monitoring, and CNCF notes that ideal candidates may already hold Kubernetes certifications such as KCNA, CKA, or CKAD, or have completed Prometheus-specific training or cloud engineer bootcamps. The credential confirms that a candidate can build or scrape observability data in an application stack whether or not it is cloud native, and it prepares candidates to work with the fundamentals of data monitoring, metrics, alerts, and dashboards using Prometheus. The exam is delivered online, is proctored, uses multiple-choice questions, and runs 90 minutes. The PCA case studies are 5 cases and about 3,227 words.
1Latency SLO hides a dependency failure
A payments platform runs 180 services across three Kubernetes clusters and 12,000 pods. The team monitors availability with a per-pod dashboard that averages CPU and request latency across all pods.
669 words, about 3 min2Reload failure after a scrape-interval override
One Prometheus server scrapes 3,800 targets across three jobs. The global scrape interval is 30s, but the api job overrides it with scrape interval: 10s and scrape timeout: 15s.
648 words, about 3 min3Histogram and error-ratio queries that look plausible
The checkout service runs on 40 pods and exports http requests total{service,pod,status code} and http request duration seconds bucket{service,pod,le}. Scrapes happen every 30s.
621 words, about 3 min4Pushgateway staleness and a histogram cardinality blow-up
A data platform runs 600 cron jobs. Each job pushes completion metrics to one shared Pushgateway with labels job, instance, and schedule.
653 words, about 3 min5Switch failure storm and a slow instance-variable dashboard
A 300-node Kubernetes cluster uses one Alertmanager route with group by: [alertname], group wait: 30s, group interval: 5m, and repeat interval: 4h. When a top-of-rack switch fails, 60 NodeDown alerts fire.
636 words, about 3 min