GCP Professional Cloud Architect Guide: How to Pass the PCA Exam
The GCP Professional Cloud Architect (PCA) is the most respected and widely cited GCP certification. It validates that you can design, implement, and manage robust, scalable, and secure cloud architectures on Google Cloud — not just operate existing ones.
Where the Associate Cloud Engineer tests operational knowledge (how to deploy and configure), the Professional Cloud Architect tests architectural judgment (which solution to design and why). The shift in emphasis is significant, and candidates who pass the ACE without real architectural experience often find the PCA exam substantially harder.
What the PCA exam tests#
The PCA uses a case study format that distinguishes it from most other cloud certification exams. Two fictional company case studies are provided in the exam — you can read them in advance on the Google Cloud website, and multiple questions in the exam reference these scenarios.
This format rewards candidates who understand architectural trade-offs at a system level, not just which GCP service does what.
Exam domains and weightings:
| Domain | Approximate weighting |
|---|---|
| Design and plan a cloud solution architecture | 24% |
| Manage and provision the cloud solution infrastructure | 15% |
| Design for security and compliance | 20% |
| Analyse and optimise technical and business processes | 18% |
| Manage implementation | 11% |
| Ensure solution and operations reliability | 12% |
Format: Approximately 60 questions, 2 hours, approximately $200, 2-year validity.
The case study format: what you need to know#
At the time of preparation, Google publishes the case studies for the PCA exam on their certification page. Read them carefully — multiple questions in the exam are anchored to the business requirements, technical constraints, and operational goals of these companies.
A typical case study describes:
- A company’s current infrastructure (often a mix of on-premises and cloud)
- Business requirements (growth targets, cost constraints, compliance needs)
- Technical requirements (availability targets, latency requirements, data residency rules)
- A team’s existing skills and tooling
Questions then ask: given these requirements, which architecture, service, or migration approach is most appropriate?
The case study format prevents pure memorisation. You have to reason about the right answer given a specific context — which is what architects actually do.
Key topics in depth#
Architecture patterns#
- High availability: Multi-region vs multi-zone architectures, Cloud Load Balancing with backend services and health checks, regional managed instance groups, Cloud Spanner for global consistency
- Disaster recovery: RTO and RPO targets, warm standby vs active-passive vs active-active, data replication across regions
- Scalability: Horizontal vs vertical scaling, stateless application design for autoscaling, managed instance groups vs GKE autoscaling
- Microservices: Cloud Run for serverless microservices, GKE for orchestrated containers, Cloud Endpoints / Apigee for API management
Data architecture#
- Storage selection: When to use Cloud Storage vs Cloud SQL vs Cloud Spanner vs Bigtable vs Firestore — and the specific requirements that drive each choice
- Analytics: BigQuery for OLAP and data warehousing, Dataflow for stream and batch processing, Pub/Sub for event ingestion, Dataproc for Hadoop/Spark workloads
- Data residency and sovereignty: How to constrain where data lives using organisation policies and resource locations
- Database migration: Database Migration Service, pglogical replication, zero-downtime migration strategies
Security architecture#
- Identity and access management at scale: IAM conditions, service account security, Workload Identity for GKE, organisation policies for guardrails
- Network security: VPC Service Controls, Private Google Access, Private Service Connect, Cloud Armour (WAF and DDoS protection)
- Encryption: Customer-managed encryption keys (CMEK), Cloud HSM, Cloud External Key Manager
- Compliance frameworks: PCI DSS, HIPAA, GDPR considerations on GCP — what GCP provides vs what the customer is responsible for
Networking architecture#
- Multi-region and hybrid connectivity: Cloud Interconnect (Dedicated and Partner), Cloud VPN, HA VPN
- Network design: Shared VPC for multi-project environments, VPC peering, hierarchical firewall policies
- DNS architecture: Cloud DNS, private zones, forwarding zones for hybrid DNS resolution
- Load balancing patterns: Global HTTPS load balancer vs regional internal load balancer, when each applies
Migration strategies#
- Lift and shift vs re-platform vs re-architect: Understanding when each is appropriate given business constraints
- Migration tools: Migrate to VMs (formerly Velostrata), Database Migration Service, Transfer Appliance, BigQuery Data Transfer Service
- Google Cloud Adoption Framework: Risk evaluation, phases of migration
Cost optimisation#
- Committed use discounts: One-year and three-year resource commitments
- Sustained use discounts: Automatic discounts for running VMs for a significant portion of the month
- Preemptible VMs: Cost savings and appropriate use cases
- Cost allocation: Resource hierarchy, labels for cost tracking, Billing Export to BigQuery
Reliability and operations#
- SRE principles: SLOs, SLIs, error budgets — how they inform design decisions
- Cloud Monitoring: Uptime checks, alerting policies, custom dashboards
- Incident response: Runbooks, incident management processes, blameless postmortems
- Deployment strategies on GCP: Traffic splitting on App Engine and Cloud Run, rolling updates in GKE, blue/green with managed instance groups
How to prepare#
Get real architectural experience first. If you have not designed and deployed a complete GCP environment — VPC, compute, database, monitoring, IAM — you will find the PCA questions abstract and difficult to reason through. The ACE gives you operational foundation; the PCA requires you to make architectural decisions from that foundation.
Read the case studies before your exam date. Google publishes them. Study them carefully. Understand each company’s requirements and think through what architectures you would propose. When you encounter case-study questions in the exam, you will be able to reason from memory rather than rereading everything under time pressure.
Practice architectural trade-offs. The PCA tests your ability to choose between architecturally equivalent solutions based on specific constraints. Practice this explicitly: when would you use Cloud Spanner instead of Cloud SQL? When would you use Bigtable instead of BigQuery? When would Shared VPC be appropriate instead of VPC peering? Understanding the why is more useful than memorising the what.
Design reference architectures. Google provides reference architectures for common patterns (web application, data lake, microservices platform, hybrid connectivity). Work through these and understand why each design decision was made.
Preparation time:
| Starting point | Typical preparation time |
|---|---|
| Holds ACE, 1+ year of GCP experience | 8–12 weeks of focused study |
| Holds ACE, limited hands-on experience | 4–6 months |
| Coming from AWS/Azure, new to GCP | Get ACE first; then return to PCA |
What this certification signals to employers#
The PCA is the GCP credential that hiring managers and technical leads look for when evaluating senior GCP engineers and cloud architects. It is more selective than the ACE — fewer engineers hold it, and the case-study format means passing it requires genuine understanding rather than study-heavy memorisation.
For engineers targeting:
- Senior cloud engineer roles at GCP-heavy organisations
- Cloud architect positions
- Technical consulting or solutions engineering at companies serving GCP customers
…the PCA is the credential to hold.
See GCP certification salary impact for how this affects earning potential.
Summary#
- The PCA tests architectural judgment, not just operational knowledge — it requires real design experience
- The case study format is distinctive — read the published case studies before your exam date
- Key areas: HA and DR design, data architecture, security at scale, migration strategies, cost optimisation
- Holds ACE first and gain real hands-on GCP experience before attempting the PCA
- 8–12 weeks of focused study is realistic for engineers with solid GCP experience