Study guide · IT & Cloud

Google Cloud Associate Cloud Engineer (ACE): Study Guide

intermediate

A practical, step-by-step plan to take GCP ACE from "interested" to exam-ready - the mechanics, what to study in what order, how to practise, and how to know you are ready.

By The Exam Atlas Editorial Team · Verified 2026-06-07

Study plans by timeline

4-week intensiveWith cloud experience (~12 hrs/week): work hands-on in Google Cloud Skills Boost, focus on compute, storage, networking and IAM.
8-week balancedThe default (~6-8 hrs/week): build console familiarity steadily, then practice questions and mocks.
12-week steadyFor those newer to GCP (~4-5 hrs/week): start with foundational concepts before hands-on labs.

What to study, in order

Weeks 1–2Setting up: projects, billing, IAM, and the gcloud CLI and Cloud Shell
Weeks 3–5Compute and storage: Compute Engine, GKE, Cloud Run, App Engine, Cloud Storage
Weeks 6–7Networking, databases and operations (monitoring, logging)
Weeks 8Access and security (IAM, service accounts); then timed reviews

The Associate Cloud Engineer is a practical exam, not a recall test. It checks whether you can actually deploy, operate, and manage workloads on Google Cloud using both the console and the gcloud command line, and it rewards the muscle memory you only get from doing the tasks. The fastest, most reliable preparation is therefore hands-on time in a free Google Cloud account and Cloud Shell, with study built around the official exam guide. This guide is a full, self-study course: it works through each of the exam’s task areas in depth, explains the trade-offs the questions hinge on (especially the four compute options and the identity model), and then turns all of it into a week-by-week plan and a final-preparation routine. It is original teaching material and study guidance only. It contains no exam questions of any kind, and you should always confirm the current exam guide against Google Cloud’s own certification page before you book, because Google revises the guide periodically.

Chapter 1: Exam overview and how to use this guide

What the Associate Cloud Engineer measures

The certification validates that you can carry out the day-to-day work of a cloud engineer on Google Cloud: setting up an environment, planning and deploying solutions across compute and storage, keeping them running, and configuring access and security. Google recommends roughly six months of hands-on experience with Google Cloud before attempting it, which is the clearest signal of its nature, this is an exam about doing, and the recommended experience is a substitute for nothing. The exam guide organises the content into four task areas: setting up a cloud solution environment, planning and implementing a cloud solution, ensuring the successful operation of a cloud solution, and configuring access and security. Google does not publish a percentage weight for each area, so treat them as roughly comparable and do not gamble on skipping one.

The format is 50 to 60 multiple-choice and multiple-select questions in two hours, delivered either online-proctored or at a test centre through Google’s testing partner. The fee is US$125. Google does not publish a numeric passing score; the result is reported as pass or fail. The credential is valid for three years, and there is a shorter renewal exam (around 20 questions in one hour) when the time comes. Because the guide is updated from time to time, always read the current version before you study in earnest, the task areas are stable, but the specific services and emphases evolve.

Where GCP fits, honestly

It is worth being clear-eyed about the credential’s value before you invest weeks in it. Google Cloud has a smaller market share than AWS and Azure, so demand is real but more concentrated, often in organisations that lean toward data, analytics, and machine learning. If your target employers run on Google Cloud, this is a strong, practical, reasonably priced entry credential and the natural foundation for the Professional Cloud Architect path. If you have no platform preference, be honest that an AWS or Azure associate certification will usually open more doors in raw job volume, so check the live demand in your region and industry rather than assuming the three clouds are interchangeable on a CV. None of that changes how you study; it just helps you decide whether to.

How to use this course

Read the chapters in order at least once. Chapter 2’s resource hierarchy and identity model underpin everything that follows, and the compute chapter is the one most worth re-reading until the four options are second nature. Treat the bold concept names as a checklist, but treat the command line as the real test: for every concept, you should be able to perform the task in Cloud Shell, not just describe it. Wherever a concept is easy to misread, a short worked illustration makes it concrete; these are teaching examples, not exam items. The final chapters convert the content into a schedule, a final-preparation routine, and a description of exam day.

Chapter 2: Setting up a cloud solution environment

This task area is the foundation: the organising structures, billing, identity, and tooling that everything else sits on. Get it solid first, because compute, storage, and operations all assume you understand projects, IAM, and the CLI.

The resource hierarchy

Google Cloud organises everything in a resource hierarchy: an organisation at the top, then folders, then projects, then the individual resources (virtual machines, buckets, and so on). The project is the fundamental unit, it groups resources, and it is the level at which billing and most permissions attach. The reason the hierarchy matters for the exam, and for real work, is that policies set higher up inherit downward: an IAM permission or an organisation policy applied at the folder level flows to the projects and resources beneath it. This is what lets a large organisation govern many projects consistently. As a teaching example: granting a team lead a role at the folder level gives them that access across every project in the folder, which is both powerful and a reason to be careful about where you attach permissions.

Billing and projects

A billing account pays for usage and is linked to one or more projects; understanding that link, and that costs are tracked per project, is the basis of cost management on Google Cloud. The exam expects you to know how to create and manage projects, associate them with billing, and use tools like budgets and alerts to keep spending visible. Practise creating a project, linking billing, and setting a budget alert, because cost awareness threads through the “planning” task area too, where you choose services partly on cost.

Identity and access management

IAM controls who (an identity) can do what (a role) on which resource. This is the single most important concept in the whole exam because it appears in every task area, so understand it precisely. Identities include human users, groups, and service accounts (covered in full in Chapter 5). Permissions are bundled into roles, of which there are three kinds: basic roles (the old, broad owner/editor/viewer roles, which are too coarse for most real use), predefined roles (curated by Google for specific services and tasks, the right default), and custom roles (which you build from individual permissions when no predefined role fits). The principle the exam rewards everywhere is least privilege: grant the narrowest role that lets someone do their job, prefer predefined over basic, and reach for custom roles only when necessary. As a teaching example of the instinct: if a user only needs to read objects in one storage bucket, you grant a predefined storage object viewer role scoped to that bucket, not the basic editor role on the whole project.

The gcloud CLI and Cloud Shell

This is the most command-line-heavy of the associate cloud exams, so the tooling is not a footnote. The gcloud CLI is the command-line interface to Google Cloud, and Cloud Shell is a browser-based terminal that comes pre-authenticated with the tools installed, the easiest place to practise. You should be comfortable configuring the CLI (setting the active project and account), and you should know the related tools: gsutil (now largely folded into gcloud storage) for Cloud Storage and kubectl for Kubernetes. The exam repeatedly rewards recognising the right command and flag, and that recognition only comes from typing them, so do the setup tasks in Cloud Shell rather than only reading about them.

Chapter 3: Compute - the four options you must know cold

Choosing the right compute service is the single most-confused area on the exam, and questions repeatedly frame a workload and ask which option fits. This chapter is the one to over-learn. The goal is to be able to say, in a sentence each, what the four options are for and when one beats the others.

Compute Engine

Compute Engine provides virtual machines (infrastructure as a service). You choose the machine type, operating system, and disks, and you manage the VM much as you would a server. You reach for Compute Engine when you need full control over the operating system and environment, when you are lifting an existing application onto the cloud with minimal change, or when a workload does not fit a more managed model. Know the cost-saving and scaling features the exam likes: managed instance groups (which let a set of identical VMs scale and self-heal), autoscaling, and pricing options such as preemptible/Spot VMs for cheap, interruptible work and committed-use discounts for steady workloads.

Google Kubernetes Engine

Google Kubernetes Engine (GKE) is managed Kubernetes for running containerised workloads with orchestration: scheduling containers across a cluster, scaling them, and recovering from failures. You choose GKE when you have containers and need the power and portability of Kubernetes, for example a microservices application with many services that must scale independently. The distinction the exam wants is GKE versus the simpler container option below: GKE gives you the full Kubernetes model and control, at the cost of more to manage.

Cloud Run

Cloud Run runs containers serverlessly: you give it a container image and it runs and scales it automatically, including scaling to zero when there is no traffic, so you pay nothing while idle. You choose Cloud Run for stateless, request-driven workloads (an API, a web service, an event handler) where you want containers without operating a Kubernetes cluster. The clean way to separate it from GKE: Cloud Run is the right answer when the scenario stresses serverless, minimal operations, scale to zero, and a single stateless container or service; GKE is the right answer when the scenario stresses full Kubernetes control or complex, multi-service orchestration.

App Engine, and choosing between them

App Engine is a managed platform as a service: you deploy application code and Google runs it without you managing servers, with the platform handling scaling. It suits web and mobile back ends where you want to focus on code. Pulling the four together, the exam’s pattern is to describe a workload and expect you to match it: full OS control or a lift-and-shift VM points to Compute Engine; containers needing Kubernetes orchestration point to GKE; a stateless container that should scale to zero with minimal ops points to Cloud Run; deploying app code on a managed platform points to App Engine. As a teaching example of reading the clues: a workload described as “a containerised stateless API that gets sporadic traffic and should cost nothing when idle” is pointing squarely at Cloud Run, because “stateless, container, sporadic, nothing when idle” are its signature.

Chapter 4: Storage, databases, and networking

Beyond compute, the “planning and implementing” task area expects you to choose and configure storage, databases, and basic networking for a workload. As with compute, the exam is about matching the option to the requirement.

Cloud Storage and storage classes

Cloud Storage is object storage for unstructured data (files, images, backups), organised into buckets. The concept the exam tests is storage classes, which trade storage cost against access frequency: Standard for frequently accessed data, Nearline for data accessed roughly monthly, Coldline for quarterly, and Archive for rarely accessed, long-term retention. The cheaper classes cost less to store but more to retrieve, so the right choice depends on how often you read the data. Know also that lifecycle rules can move objects to colder classes automatically as they age. As a teaching example: logs you must keep for compliance but will almost never read belong in Archive, while images served constantly on a website belong in Standard.

Choosing a database

Google Cloud offers several managed databases, and the exam expects you to pick sensibly rather than know every detail. Cloud SQL is managed relational (MySQL, PostgreSQL, SQL Server) for traditional applications that need a standard relational database. Firestore is a document (NoSQL) database for flexible application data. Bigtable is a wide-column NoSQL store for very large analytical and operational workloads. Spanner is a globally distributed, strongly consistent relational database for large-scale, mission-critical systems. BigQuery is a serverless data warehouse for analytics over large datasets using SQL. The matching instinct: a standard app database points to Cloud SQL; massive analytical queries point to BigQuery; huge-scale NoSQL points to Bigtable; flexible app documents point to Firestore; global relational scale points to Spanner.

Networking basics

Networking on the exam is foundational rather than deep, but you must understand the Virtual Private Cloud (VPC), which is Google Cloud’s software-defined network and is global by design (a single VPC can span regions, with subnets in each). Know subnets, firewall rules (which control traffic to and from resources), and at a high level Cloud Load Balancing for distributing traffic. The point to carry into the exam is that the VPC is the network backbone everything connects to, and firewall rules are how you control access at the network layer, complementing IAM at the identity layer.

Chapter 5: Operations, access, and security

The remaining two task areas, ensuring successful operation and configuring access and security, are where running systems are kept healthy and access is kept tight. Both reward the same instinct: visibility and least privilege.

Monitoring, logging, and operations

Keeping a solution running well means being able to see what it is doing. Google Cloud’s Cloud Operations suite provides Cloud Monitoring (collecting metrics and building dashboards and alerts) and Cloud Logging (storing, searching, and analysing logs). The exam expects you to know how to observe resources, set up basic alerting so you are told when something is wrong, and manage and maintain running resources over time. The mindset is proactive: you instrument and monitor so problems surface early, rather than discovering an outage from users. Practise viewing logs and creating a simple alerting policy, because “how would you know if X happened?” is a recurring shape of operations questions.

Service accounts and workload identity

Chapter 2 introduced IAM for human users; this is the other half. A service account is a special identity used by applications and virtual machines (not people) to authenticate to Google Cloud services. When a VM or a Cloud Run service needs to read a storage bucket, it does so as a service account with an appropriate role. The exam tests whether you understand that workloads get their own identities and least-privilege roles, just as people do, and that you should attach a narrowly scoped service account to a workload rather than over-granting. As a teaching example: an application that writes to one bucket should run as a service account holding only the storage object creator role on that bucket, so a compromise of the app cannot reach anything else.

Security as least privilege everywhere

Pulling the security thread together, the exam’s consistent expectation is least privilege applied through IAM at the identity layer and firewall rules at the network layer. Prefer predefined roles over basic ones, scope permissions to the narrowest resource that works, give workloads their own service accounts, and control network access with firewall rules on the VPC. Most “which approach is most secure?” questions resolve to the option that grants the least access needed to accomplish the task, so when two answers both work, prefer the more tightly scoped one.

Chapter 6: Study plan and timeline

With the content understood, the remaining work is pacing it and, above all, making sure the hands-on practice does not get squeezed out. Two things drive the plan: confirming the current exam guide, and spending your hours in the console and CLI rather than only reading.

Confirm the guide, then learn by doing

Start by reading the current exam guide on Google Cloud’s certification page, since it is revised periodically and defines your scope. Then commit to learning by doing in a free Google Cloud account and Cloud Shell. For every task area, perform the real tasks both in the console and with the equivalent gcloud command: create a project and set billing, assign IAM roles and a service account, launch a Compute Engine instance, deploy something to Cloud Run or GKE, create a storage bucket, and view logs and an alert. The exam rewards recognising the right command and flag, and that recognition is almost impossible to build from notes alone.

An eight-week default, and how to flex it

A balanced plan runs about eight weeks at six to eight hours a week. Weeks one to two cover setting up: projects, billing, the resource hierarchy, IAM, and getting comfortable in the gcloud CLI and Cloud Shell. Weeks three to five cover compute and storage: the four compute options (with extra time, because this is the most-confused area), plus Cloud Storage and the managed databases. Weeks six to seven cover networking, and operations: the VPC and firewall rules, then Cloud Monitoring and Logging. Week eight covers access and security consolidation (IAM roles and service accounts) and then timed practice. If you already have cloud experience, compress to a roughly four-week intensive at around twelve hours a week, working hands-on and concentrating on compute, storage, networking, and IAM. If you are newer to GCP, stretch to twelve weeks at four to five hours, spending the early weeks on foundational concepts before the hands-on labs. To turn whichever timeline you choose into dated weeks for your own start date, use the free study-plan generator.

Where to spend the extra hours

Because Google does not publish task-area weights, you cannot study to a percentage, but two areas reliably repay extra time. The four compute options are the most-confused topic on the exam, so over-learn them until you can match any described workload to Compute Engine, GKE, Cloud Run, or App Engine in a sentence. And the IAM model, basic versus predefined versus custom roles, service accounts for workloads, and least privilege, runs through every task area, so make it second nature. If you are still choosing between platforms, the honest framing from Chapter 1 applies: GCP is strong where employers lean toward data and ML, and trails AWS and Azure in raw volume, so let your target employers, not the syllabus, decide whether this is the right cloud for you.

Chapter 7: Final preparation, exam day, and format

Final preparation

In the final weeks, shift from learning new services to timed practice and consolidation. Sit timed practice exams and, crucially, read each scenario for the binding requirement before choosing, the compute-choice, IAM, and networking questions are where a single word in the scenario decides the answer, so review the reasoning behind every miss rather than only noting the correct option. In the closing week, consolidate instead of adding new material: revisit the four compute options, the IAM model, and the resource hierarchy, run a final timed mock, and confirm the current exam guide on the Google Cloud site. Steer clear of sites that leak or sell stolen test content; they breach Google’s certification policy and copyright, and they do nothing to build the hands-on recognition the exam actually rewards.

Does AWS or Azure experience transfer?

Partly, and it is worth naming the trap. The underlying concepts, compute, storage, networking, identity, cost, carry over from other clouds, but Google Cloud’s terminology, defaults, and resource hierarchy differ, and assuming a one-to-one mapping is a reliable way to get questions wrong. If you come from AWS or Azure, learn Google Cloud’s own model and its gcloud commands rather than translating in your head; the exam tests GCP as GCP, not as a relabelled version of another platform.

Exam day and format

On the day, expect 50 to 60 multiple-choice and multiple-select questions in two hours, taken online-proctored or at a test centre, so prepare your environment and identification in advance. The pace is comfortable compared with shorter exams, but do not let it lull you: read each scenario carefully for the requirement that pins the answer, and apply the instincts you built over the weeks, match the workload to the right compute option, prefer least-privilege IAM, give workloads their own service accounts, and choose the storage class or database that fits the access pattern. Having practised the real tasks in the console and CLI, the questions will read like work you have already done, which is exactly the advantage the hands-on weeks were meant to build.

Key concepts to master

Resource hierarchy
Organisation → folders → projects → resources. IAM and billing attach here.
gcloud CLI and Cloud Shell
The exam expects comfort with command-line operations, not just the console.
Compute options
Compute Engine (VMs), GKE (Kubernetes), Cloud Run (containers), App Engine (PaaS) - know when to use each.
IAM and service accounts
Roles (basic, predefined, custom) and service accounts for workload identity.
Cloud Storage classes
Standard, Nearline, Coldline, Archive - cost vs access frequency.

What you should be able to do

By exam day, you should be able to:

  • Set up a GCP project, IAM and billing
  • Deploy compute with Compute Engine, GKE and Cloud Run
  • Configure storage and basic VPC networking
  • Use monitoring and logging at a basic level
  • Complete hands-on console tasks

How to practise

Practise in the console via Skills Boost, since the exam expects hands-on familiarity. Drill scenario questions and review weak areas before booking.

  • Practise actively from early on - recall and apply, don't just re-read.
  • Each week, review the previous week's weak spots before moving on.
  • Do at least one full-length, timed mock near the end, then a second after fixing weak areas.
  • Warm up with our original GCP ACE practice questions (concept checks, not exam dumps).

We never publish exam dumps or "real" questions. Use official practice and reputable providers for question banks.

Are you ready? (readiness checklist)

  • You score at or above the pass mark (Not published by Google) on full-length, timed mocks - consistently, not once.
  • No more than one or two weak domains remain, and you know exactly which.
  • You can explain why the wrong options are wrong, not just spot the right one.
  • You've completed at least one full-length mock under real time pressure.
  • You could pass next week, not only on the day you crammed.

On exam day

Online-proctored or at a test centre via Google's testing partner; around 50 multiple-choice and multiple-select questions in 2 hours.

  • Arrive early, or run the online-proctoring system check well ahead; have valid ID ready.
  • Budget your time per question and keep moving - don't sink minutes into one item.
  • Where the format allows, flag hard questions and return to them rather than stalling.
  • Read scenario and performance-based questions twice: work out what is actually asked first.
  • Taper in the final days - light review and rest beat an all-nighter.

Common mistakes to avoid

  • Studying without using gcloud; the exam is practical and CLI-heavy.
  • Confusing the compute options (Compute Engine vs GKE vs Cloud Run vs App Engine).
  • Neglecting IAM and service accounts, which appear throughout.
  • Assuming AWS/Azure knowledge maps directly; GCP terminology differs.

Resource stack

Start with the free and official resources above. Paid courses and question banks help if you want structure, but they are optional, not required to pass.

What to study next

ACE is the foundation for GCP roles; next is Professional Cloud Architect. Compare with AWS SAA if choosing a platform.

FAQ

How long does it take to study for GCP ACE?
Most people need 40–70 hours over six to ten weeks. Hands-on practice with the gcloud CLI is the fastest way to prepare.
Is GCP ACE in demand?
Demand is real but more concentrated than AWS or Azure, often in data and ML-heavy organisations. It is most valuable where employers specifically use Google Cloud.
Do I need to know the gcloud command line?
Yes. The exam expects comfort with both the console and the gcloud CLI in Cloud Shell.
How do I choose between Compute Engine, GKE, Cloud Run and App Engine?
Compute Engine gives you virtual machines when you need full control; GKE runs containers at scale with Kubernetes; Cloud Run runs stateless containers serverlessly; App Engine is a managed platform for apps. The exam tests whether you can match the workload to the right option, so learn the trade-offs, not just the names.
How many practice exams should I do for GCP ACE?
Sit several timed practice exams in the final weeks, focusing on the scenario questions about compute choice, IAM and networking. Aim to be comfortably above the pass mark on fresh questions before booking.
Does my AWS or Azure experience transfer to GCP?
Partly. The concepts - compute, storage, networking, identity, cost - carry over, but Google Cloud's terminology, defaults and the resource hierarchy differ. Do not assume a mapping; learn GCP's own model and the gcloud commands.

Sources