GCP Prep

A Cloud Certification Study Plan

GCP Prep5 min readUpdated September 2, 2026

This is a concrete eight-week plan built around five to seven hours a week. It is written for Associate Cloud Engineer, but the structure — study, build, drill, review — adapts to any cloud exam.

Table of contents

How the plan works

Each week has the same shape: a study block covering new material, a build block where you use it, and a drill block of practice questions on that topic. The pattern matters more than the specific hours.

  • Study — around 2 hours. Read or watch material on the week's topic.
  • Build — around 2 hours. Create it yourself, then deliberately break it.
  • Drill — around 1 hour. Practice questions on that topic only, reading every explanation.
  • Review — around 30 minutes. Revisit last week's weak points before moving on.

Weeks 1–2: foundations and compute

Week 1 — Platform fundamentals. Projects, folders, organisations, regions, zones, billing structure and the resource hierarchy. Install the command-line tools and get comfortable with configurations.

Build: create a project, set a budget alert, enable a few APIs, and create your first virtual machine from both the console and the command line so you can see how they correspond.

Week 2 — Compute Engine. Machine types, images, persistent disks, snapshots, instance templates, managed instance groups and autoscaling.

Build: create a custom image with a simple web server, build an instance template from it, create a managed instance group across two zones, then delete an instance and watch autohealing replace it.

Weeks 3–4: storage, networking

Week 3 — Storage. Cloud Storage classes, lifecycle rules, versioning, access control and signed URLs. Persistent disk types and when a shared file system is required instead.

Build: create a bucket, upload objects, apply a lifecycle rule, enable versioning and overwrite an object to see what happens, then generate a signed URL and wait for it to expire.

Week 4 — Networking. VPCs, subnets, routes, firewall rules and priority, and the load balancer decision tree.

Build: create a custom VPC with a private subnet, add Cloud NAT, confirm instances can reach the internet but nothing can reach them. Then put a load balancer in front of your instance group and verify the health checks.

Weeks 5–6: IAM, databases and Kubernetes

Week 5 — IAM and security. This is the highest-yield week. Basic, predefined and custom roles, policy inheritance, service accounts, impersonation and least privilege.

Build: create a service account with a single narrow role, attach it to an instance, and have an application use it. Then attempt an action it lacks permission for and read the error carefully — recognising these messages is directly examinable.

Week 6 — Databases and Kubernetes. Cloud SQL configuration and high availability, plus the selection criteria across Cloud SQL, Firestore, Bigtable and Spanner. Then a GKE cluster and basic workload management.

Build: create a Cloud SQL instance with private IP and connect securely. Create a small cluster, deploy a workload, expose it with a service, and scale it up and down.

Weeks 7–8: operations and consolidation

Week 7 — Monitoring and operations. Dashboards, alerting policies, uptime checks, log queries, log-based metrics and log sinks.

Build: create an uptime check against your load balancer, an alerting policy that notifies you, then deliberately break the backend and confirm the alert fires. Query the logs to find the failure.

Week 8 — Consolidation. No new material. Full-length timed mock exams, review of weak topics, and command-line drilling until common commands are automatic.

  1. 1Day 1 — full timed mock, no notes. Record the score and the weak topics.
  2. 2Days 2–4 — work only on the weakest topics, in the console rather than in notes.
  3. 3Day 5 — second full timed mock.
  4. 4Days 6–7 — light review, logistics check, rest.

Adapting the plan

  • For a Professional exam, extend to twelve to sixteen weeks and add weeks for the specialist domains. Keep the same weekly shape.
  • For Cloud Digital Leader, compress to three weeks and replace the build blocks with case-study reading, since the exam does not test configuration.
  • If you have less time per week, stretch to twelve weeks rather than cutting the build blocks. Hands-on practice is the part that produces retention.
  • If you already work on the platform, skip the build blocks for topics you use daily and spend the time on the ones you avoid at work — that is where your gaps are.

Frequently Asked Questions

Is eight weeks realistic?
For someone with general IT experience studying five to seven hours a week, yes. Complete beginners should plan for twelve weeks. People already working on Google Cloud daily often need four to six.
What if I miss a week?
Repeat rather than skip. Missing a week is normal; skipping content is what causes trouble later, because cloud topics build on each other more than the syllabus structure suggests.
Can I study more hours per week and finish faster?
To a point. Beyond roughly ten to twelve hours a week retention drops noticeably, and hands-on practice needs time between sessions to consolidate. Four weeks of intense cramming generally produces a weaker result than eight steady ones.
Should I do practice questions throughout or only at the end?
Throughout, on the topic you just studied. They are a diagnostic tool, and using them early tells you what you misunderstood while there is still time to fix it.

Put this into practice

Reading is the easy part. Work through original practice questions with full explanations to find out what you actually know.