I passed the GCP Professional Cloud Architect exam last month with about 8 weeks of focused study. No bootcamp. No $2,000 course. Just a structured plan and a lot of discipline.
Here's exactly what I did, week by week, including what worked and what I'd skip next time.
Before We Start: The Exam at a Glance
- Format: 50 questions, 2 hours
- Cost: $200
- Passing score: Not disclosed (Google doesn't publish it)
- Prerequisites: None officially, but Google recommends 3+ years of industry experience and 1+ year with GCP
- Case studies: 4 pre-published case studies you MUST memorize
That last point is crucial. Google publishes four fictional company case studies on their exam page. Questions will reference these companies by name and ask you to design solutions for them. If you walk in without memorizing the case studies, you're throwing away free points.
Week 1: Foundations
Goal: Understand GCP's core services and how they compare to AWS/Azure equivalents.
I watched the Google Cloud Fundamentals course on Coursera. It's dry but thorough. Key services to nail down:
- Compute Engine vs App Engine vs Cloud Run vs GKE
- Cloud Storage classes (Standard, Nearline, Coldline, Archive)
- VPC networking, subnets, firewall rules
- IAM, service accounts, organization policies
If you're coming from AWS, the mental model shift is smaller than you'd think. GCP's IAM is more hierarchical though — org → folder → project → resource. Get comfortable with that hierarchy.
Week 2: Networking Deep Dive
This is where GCP gets interesting and where a lot of people struggle.
The PCA exam is obsessed with networking. Shared VPCs, VPC peering, Cloud Interconnect vs Cloud VPN, load balancer types (there are like six different ones), and Private Google Access.
I spent the entire week on networking alone. Built a Shared VPC with two service projects. Set up an HTTP(S) load balancer with managed instance groups. Configured Cloud NAT.
Pro tip: Know the difference between Premium and Standard network tiers. This shows up on the exam more than you'd expect.
Week 3: Data and Storage
BigQuery is the star here. Know the difference between federated queries and native tables. Understand partitioning (time-based and integer-range) and clustering. Know when to use BigQuery ML vs Vertex AI.
Also cover:
- Cloud SQL vs Cloud Spanner vs AlloyDB (when to use each)
- Dataflow for streaming and batch processing
- Pub/Sub as the messaging backbone
- Dataproc for Hadoop/Spark workloads (and when to migrate to Dataflow instead)
Week 4: Security and Compliance
IAM policies, VPC Service Controls, Cloud KMS, Secret Manager, Binary Authorization for GKE.
The exam tests whether you can design a security architecture that meets compliance requirements. You'll get scenarios like "a healthcare company needs to process patient data" — and you need to know about VPC Service Controls, CMEK, and data residency.
Don't skip this week. Security questions are scattered across every section of the exam.
Week 5: Case Studies
Stop everything else. Read all four case studies. Then read them again.
For each case study, make a one-page cheat sheet:
- Current architecture (on-prem, existing cloud, hybrid)
- Business requirements (cost, speed, compliance)
- Technical requirements (latency, availability, specific services)
- Pain points (what's broken or inefficient)
Then, for each case study, write down your proposed GCP architecture. What services would you use? How would you handle migration? What's the network topology?
When exam questions reference these companies, you should already have the answers in your head.
Week 6-7: Practice Exams
This is where the exam is won or lost.
I used GCP PCA practice questions from ExamCert — $4.99 for lifetime access to hundreds of architect-level questions. The explanations were genuinely helpful, not just "the answer is B." They explain why B is right and why A, C, and D are wrong in the specific context.
Compared to Whizlabs ($20-30 for limited question sets) or the $300+ bootcamp practice exams, ExamCert was the obvious choice. Plus the money-back guarantee if you don't pass removes all the risk.
I took a practice exam every other day. Reviewed wrong answers. Took notes on patterns I kept missing. My progression:
- First attempt: 58% (humbling)
- Second: 67%
- Third: 74%
- Fifth: 82%
- Final: 88%
Week 8: Final Review + Exam
Reviewed case studies one more time. Re-read my notes on networking (my weakest area). Did one final timed practice exam — scored 85%.
Booked the exam for Thursday morning. Fresh mind, not too early.
Exam Day Notes
- I finished in about 90 minutes. Don't rush, but 2 hours is generous for 50 questions.
- About 8-10 questions directly referenced case studies
- Networking was easily 25-30% of my exam
- Several questions about migrating from on-prem to GCP
- Two questions about cost optimization that had tricky answers
What I'd Skip Next Time
Honestly? I over-studied Kubernetes. The PCA exam asks about GKE at a high level — when to use it, how to configure node pools, Autopilot vs Standard. It doesn't ask you to write YAML manifests.
The Bottom Line
8 weeks. $200 exam fee. $4.99 for practice exams. Maybe $0-30 for online courses.
The GCP PCA is one of the highest-respected cloud architecture certs in the industry. Start with a free GCP PCA practice test to see where you stand, then follow the plan.
You've got this.
Top comments (0)