
PCA (Prometheus Certified Associate) is the Linux Foundation and CNCF's beginner certification for Prometheus, the most widely used open-source metrics and alerting system in Kubernetes and DevOps teams. It covers observability concepts, how Prometheus scrapes and stores metrics, PromQL (the largest domain), instrumentation and exporters, and alerting with Alertmanager and dashboards. For DevOps and SRE freshers in India, monitoring skills come up in almost every interview, and PCA is a structured way to prove them.
This guide follows the official Linux Foundation exam page: the five domains with weights, the PromQL you must be able to read, a 3-week lab plan and 15 original practice questions.
Quick facts (Linux Foundation, November 2026)
| Item | Detail |
|---|---|
| Format | Online, proctored, multiple choice |
| Duration | 90 minutes |
| Level | Beginner; no prerequisites |
| Price | US$250 exam only; US$299 with the LFS241 course; US$495 with an annual training subscription |
| Validity | 2 years |
| Eligibility | 12 months to schedule and take the exam; one retake included |
Read the official PCA exam page and candidate handbook before booking.
Domains and weights
| Domain | Weight | Sub-topics (from the Linux Foundation) |
|---|---|---|
| PromQL | 28% | Selecting data; rates and derivatives; aggregating over time; aggregating over dimensions; binary operators; histograms; timestamp metrics |
| Prometheus Fundamentals | 20% | System architecture; configuration and scraping; Prometheus limitations; data model and labels; exposition format |
| Observability Concepts | 18% | Metrics; logs and events; tracing and spans; push vs pull; service discovery; SLOs, SLAs and SLIs |
| Alerting and Dashboarding | 18% | Dashboarding basics; alerting rules; Alertmanager; when, what and why to alert |
| Instrumentation and Exporters | 16% | Client libraries; instrumentation; exporters; structuring and naming metrics |
Prometheus in five facts
- Pull model: Prometheus scrapes HTTP
/metricsendpoints on a schedule. Short-lived batch jobs that can't be scraped push to the Pushgateway instead. - Data model: every time series is a metric name plus key-value labels. Each new label value creates a new series, so never put user IDs or request IDs in labels (high cardinality).
- Four metric types: counter (only goes up, such as requests served), gauge (goes up and down, such as memory in use), histogram (bucketed observations, such as request latency) and summary (client-side quantiles).
- Service discovery: targets come from static configs or discovery for Kubernetes, cloud providers, DNS and more.
- Limitations: Prometheus is for metrics, not logs or exact per-event data (such as billing). Local storage is single-node; long-term, highly available storage uses remote write to systems such as Thanos, Cortex or Mimir.
PromQL you must be able to read
# Per-second request rate over the last 5 minutes, per service
sum by (service) (rate(http_requests_total[5m]))
# Error ratio (5xx as a share of all requests)
sum(rate(http_requests_total{status=~"5.."}[5m]))
/ sum(rate(http_requests_total[5m]))
# 95th percentile latency from a histogram
histogram_quantile(0.95,
sum by (le) (rate(http_request_duration_seconds_bucket[5m])))
# Targets that are down
up == 0
# Memory used by each pod, top 5
topk(5, sum by (pod) (container_memory_working_set_bytes))
Remember: use rate() on counters, never on gauges; use increase() for "how many in the last hour"; an instant vector has one value per series, a range vector ([5m]) has many.
An alerting rule
groups:
- name: web
rules:
- alert: HighErrorRate
expr: |
sum(rate(http_requests_total{status=~"5.."}[5m]))
/ sum(rate(http_requests_total[5m])) > 0.05
for: 10m
labels:
severity: page
annotations:
summary: "More than 5% of requests are failing"
Prometheus evaluates the rule; Alertmanager groups, deduplicates, silences, inhibits and routes the notification to email, Slack, PagerDuty and so on. Alert on symptoms users feel (errors, latency, SLO burn), not on every CPU spike.
3-week lab plan
| Week | Focus | Lab |
|---|---|---|
| 1 | Observability concepts, SLIs and SLOs, architecture, scrape config, data model, exposition format | Run Prometheus and node_exporter in Docker; add a scrape job |
| 2 | PromQL: selectors, rate, aggregation, binary operators, histograms | Write 30 queries against node_exporter and a demo app |
| 3 | Instrumentation with a client library, exporters, metric naming; alerting rules, Alertmanager, Grafana dashboards | Instrument a small Python app; route one alert to email; book the exam |
15 practice questions with answers
These are original questions written for this guide, not exam questions.
1. Does Prometheus push or pull metrics by default?
Pull: it scrapes targets over HTTP.
2. A nightly batch job finishes in 30 seconds. How do its metrics reach Prometheus?
Push them to the Pushgateway, which Prometheus scrapes.
3. Which metric type fits "number of orders processed since start"?
A counter.
4. Which metric type fits "current queue length"?
A gauge.
5. Why is adding a user_id label a bad idea?
It creates a new time series per user (high cardinality), which overloads Prometheus.
6. Which function gives a per-second average rate of a counter?
rate().
7. Which function calculates a percentile from histogram buckets?
histogram_quantile().
8. What does the expression up == 0 return?
Targets that Prometheus failed to scrape.
9. Which operator keeps only the series with the label service="cart"?
A label matcher in the selector: metric{service="cart"}.
10. What is the difference between sum by (pod) and sum without (pod)?
by keeps only the pod label; without removes the pod label and keeps the rest.
11. What does the for: 10m clause in an alerting rule do?
The condition must stay true for 10 minutes before the alert fires.
12. Which component silences or routes alerts to Slack?
Alertmanager.
13. You need to monitor a MySQL server that has no /metrics endpoint. What do you use?
An exporter (for example mysqld_exporter).
14. What is the naming convention for a counter of HTTP requests?
snake_case with a unit or _total suffix, such as http_requests_total.
15. You must keep metrics for two years across several clusters. What do you add?
Remote write to a long-term store such as Thanos, Cortex or Mimir.
After PCA
- Learn monitoring as part of our SRE course and Kubernetes course.
- Add KCNA or CKAD, and the OpenTelemetry OTCA for traces and logs.
- See live DevOps and SRE jobs.
Cloud Soft Solutions in Ameerpet, Hyderabad teaches DevOps, Kubernetes, SRE and observability with real-time projects. See the APEX program and all courses.
📞 Call +91 96660 19191 · WhatsApp · Flat No. 513, 5th Floor, Nilagiri Block, Aditya Enclave, Ameerpet, Hyderabad
Frequently Asked Questions
What is the Prometheus Certified Associate exam?
A Linux Foundation and CNCF online, proctored, multiple-choice exam on Prometheus, PromQL, instrumentation and alerting.
How long is the PCA exam?
90 minutes.
Does PCA have prerequisites?
No. It is a beginner-level exam with no prerequisites.
Which PCA domain has the most weight?
PromQL at 28%.
How long is the PCA certification valid?
2 years.
Is a retake included?
Yes. The official page lists one retake.



