Google Cloud

What's Actually on the GCP Professional Cloud Architect Exam (and How to Pass It)

The GCP PCA is one of the most respected cloud architecture certifications available. It is also genuinely hard. This guide covers exactly what's tested, the case studies you must study, and a realistic preparation strategy.

CertifySphere Team
March 20, 2026
12 min read

The Google Cloud Professional Cloud Architect (PCA) certification is not a memorization exam. Google is explicit about this: the exam tests your ability to evaluate architectural trade-offs, make design decisions under constraints, and apply GCP services to realistic business scenarios. You cannot pass it by reading flashcards.

Exam Structure at a Glance

Questions50–60 multiple choice and multiple select
Duration120 minutes
Passing scoreApproximately 70% (not published)
FormatRemote or at a testing center
Cost$200 USD
Validity2 years

Domain Breakdown

Designing Cloud Solutions (26%)

Choosing the right services for compute, storage, networking, and data management given functional and non-functional requirements.

Managing & Provisioning Infrastructure (20%)

Automating infrastructure with Terraform, Deployment Manager, and Cloud Build. Configuring billing, quotas, and resource hierarchy.

Designing for Security & Compliance (20%)

IAM, VPC Service Controls, data encryption, identity federation, and regulatory compliance patterns.

Analyzing & Optimizing Solutions (18%)

Cost management, performance optimization, capacity planning, and Cloud Monitoring/Logging.

Managing Implementation (16%)

Managing development and operations processes, ensuring quality, change management, and migration strategies.

The Case Studies: Your Most Important Study Material

The PCA exam includes scenario-based questions drawn from four published case studies. Google makes these publicly available before the exam. You should read each one multiple times and be able to answer architectural questions about them without re-reading.

Advertisement

Mountkirk Games

Online gaming company migrating from a colocation data center to GCP. Needs global scale, low latency, real-time analytics, and rapid development velocity.

Key exam focus: Focus on GKE for game backend, Cloud Spanner for global transactional data, Cloud Pub/Sub for event ingestion, and BigQuery for analytics.

Dress4Win

Online fashion retailer with a hybrid setup: some workloads on-prem, some on GCP. Needs to reduce latency, improve DR posture, and modernize legacy applications.

Key exam focus: Focus on hybrid connectivity (Cloud Interconnect vs. VPN), Compute Engine lift-and-shift vs. containerization trade-offs, and Managed Instance Groups for autoscaling.

TerramEarth

Heavy equipment manufacturer connecting industrial machines via IoT. Needs to process massive streams of telemetry data for predictive maintenance.

Key exam focus: Focus on Cloud IoT Core, Pub/Sub, Dataflow (streaming), BigQuery (analytics), and Vertex AI for ML-based predictions.

EHR Healthcare

Healthcare records company moving to GCP with strict compliance requirements (HIPAA). Needs multi-regional availability, data security, and audit trails.

Key exam focus: Focus on VPC Service Controls, CMEK (customer-managed encryption keys), Cloud Audit Logs, Cloud Healthcare API, and compliance-specific IAM patterns.

GCP Services You Must Know Deeply

The PCA is breadth-first but tests certain services at depth. Prioritize these:

Compute Engine + MIGs + Autoscaling
GKE (Standard vs. Autopilot)
Cloud Run + App Engine
Cloud Storage (classes, lifecycle)
Cloud Spanner vs. Cloud SQL vs. Firestore
BigQuery (partitioning, clustering, DML)
Pub/Sub + Dataflow (streaming)
Cloud Interconnect vs. Cloud VPN
Shared VPC + VPC peering
VPC Service Controls
Cloud IAM (org policies, conditions)
Cloud Monitoring + Alerting

How the Exam Tests Architectural Judgment

Unlike the AWS SAA, the GCP PCA rarely has "obviously wrong" answers. Most questions present four architecturally valid options that differ in cost, reliability, latency, complexity, or compliance characteristics. The exam is testing whether you understand the trade-offs, not just what the services do.

Examples of the kinds of judgment calls tested:

  • A company needs a globally distributed database with strong consistency. Do you use Cloud Spanner, Cloud SQL with read replicas, Firestore, or Bigtable? (Cloud Spanner — the only option that gives global strong consistency.)
  • An application needs to process 500,000 events per second with exactly-once semantics. Do you use Pub/Sub alone, Pub/Sub + Dataflow, Kafka on GKE, or Cloud Composer? (Pub/Sub + Dataflow — Dataflow provides exactly-once processing; Pub/Sub alone doesn't.)
  • A healthcare workload needs HIPAA compliance with audit trails. Which logging and monitoring configuration is correct? (Cloud Audit Logs with Data Access logs enabled, exported to Cloud Storage for long-term retention.)

Realistic Study Plan (8–10 Weeks)

The PCA is harder than the AWS SAA and requires more preparation time for most candidates.

  • Weeks 1–2: GCP fundamentals. Complete the Google Cloud Digital Leader or ACE (Associate Cloud Engineer) material if you are new to GCP. The PCA assumes you already know the basics.
  • Weeks 3–5: Core services deep dive. Cover compute, storage, networking, and databases. For each category, understand when to use each service and why. Do labs on Google Cloud Skills Boost.
  • Week 6: Case studies. Read all four case studies twice. For each one, make a list of: (a) the company's requirements, (b) the constraints, and (c) which GCP services are the natural fit and why.
  • Weeks 7–8: Security and compliance. This is 20% of the exam. Study IAM deeply, VPC Service Controls, encryption options, and compliance-relevant services (Cloud Healthcare API, Assured Workloads).
  • Weeks 9–10: Practice exams and review. Take full practice exams. For every wrong answer, trace back to the service, the trade-off, or the case study you misunderstood.

Common Mistakes on the GCP PCA

Treating it like a memorization exam — it tests judgment, not recall
Skipping the case studies — roughly 20–25% of questions reference them
Confusing Cloud SQL (regional managed PostgreSQL/MySQL) with Cloud Spanner (global, horizontally scalable) — they are not interchangeable
Underestimating networking — shared VPC, VPC peering, and connectivity options appear frequently
Not practicing architecture decisions: practice by designing systems, not just by reading about them

Practice for the GCP PCA

Take a free timed mock exam for the GCP Professional Cloud Architect on CertifySphere.

Start free mock exam →