Domains and weights from Google Cloud's exam guide.
Domain
Weight
Setting up a cloud solution environment
20%
Planning and configuring a cloud solution
17%
Deploying and implementing a cloud solution
25%
Ensuring successful operation of a cloud solution
20%
Configuring access and security
18%
Free ACE sample questions
The first questions of the free diagnostic. Pick your answer, then check it. Why every option is right or wrong, and the What-If twist, open in the workspace.
Question 1 · Deploying and implementing a cloud solution
A colleague created a cluster from their workstation, and you must inspect it from a fresh Cloud Shell session without changing the cluster or its node pools in any way. Your account has the Kubernetes Engine Developer role. Here is the session:
Bash
gcloud container clusters list
# NAME LOCATION STATUS
# shop-cluster us-east1 RUNNING
kubectl get namespaces
# error: no configuration has been provided,
# try setting KUBERNETES_MASTER environment variable
Question 2 · Setting up a cloud solution environment
A developer with the Editor role deploys the first service in a brand-new project, and the release must go out today without creating another project or asking for broader roles. The command and its error are:
Bash
gcloud run deploy api \
--image=us-docker.pkg.dev/acme-prod/web/api:1.4 \
--region=us-east1 --project=acme-new
# ERROR: (gcloud.run.deploy) PERMISSION_DENIED:
# Cloud Run Admin API has not been used in project acme-new
# before or it is disabled.
What is the root cause, and which step fixes it?
AThe project is broken, so create a new project and deploy there
BA Cloud Run quota is exhausted, so request a quota increase
CThe developer lacks rights, so ask for the Owner role on the project
DThe API is off in this project, so run gcloud services enable run.googleapis.com
Check the answer
Answer: D. Most Google Cloud APIs are disabled in a new project; enable a service with gcloud services enable before using it.
An image-processing application on a Compute Engine VM must read objects from a Cloud Storage bucket in the same project. Security forbids downloadable credentials on any VM, and the application uses Google client libraries that look for Application Default Credentials. Given these limits, what should the cloud engineer do so that the code works without a key file?
AConfigure Workload Identity Federation for the VM
BCreate a service account key and copy it to the VM
CAttach a service account with the Storage Object Viewer role to the VM
DSign in on the VM with a developer's user account
Check the answer
Answer: C. Workloads on Compute Engine authenticate with the VM's attached service account, with no key.
Resource hierarchy, organization policies, projects, APIs and quotas
IAM allow policies are inherited from organization to folder to project to resource, and the effective access is the union of every binding along that path.
Organization policy constraints restrict what configurations are possible regardless of IAM, so even a project Owner cannot create a resource that a constraint forbids.
Folders group projects by department or environment, which lets one role binding or one organization policy apply to many projects without repeating it.
Each project has its own set of enabled APIs and its own quotas, so a new project needs services enabled before the first gcloud command against them works.
The project ID is chosen at creation, is globally unique and can never be changed, while the project name is a display label that can be edited freely.
Cloud Billing: accounts, budgets and alerts, billing export, labels and committed use
Linking a project to a billing account needs the Billing Account User role on the account together with billing permission on the project, which Owner or Project Billing Manager provides.
A budget only notifies: threshold rules send email and optional Pub/Sub messages, and capping spend requires automation that acts on those programmatic notifications.
Cloud Billing export to BigQuery is the source for detailed cost analysis by SKU, project and label, and it only contains data from the time export was enabled.
Labels attached to resources flow into the billing export, so teams that share a single project can still be charged back for what each one used.
Committed use discounts fit predictable always-on workloads, while Spot VMs fit fault-tolerant batch work that can restart when Compute Engine preempts an instance.
Choosing compute: Compute Engine, GKE Standard and Autopilot, Cloud Run and Cloud Run functions
Compute Engine fits workloads that need operating system access, specific machine types, licences bound to hardware or software that cannot be containerized without rework.
GKE Autopilot manages nodes, scaling and upgrades for you and bills for Pod requests, while GKE Standard exposes node pools for teams needing node-level control.
Cloud Run services scale stateless containers with request traffic, down to zero when idle, and need no cluster or VM for the team to manage.
Cloud Run jobs execute a container until it exits, which suits scheduled exports and batch tasks that do not listen for HTTP requests at all.
Spot VMs are much cheaper but can be preempted at any moment, so they suit checkpointed batch work and never a single stateful production server.
Planning networks: load balancer choice, Network Service Tiers, Cloud DNS and hybrid connectivity
A global external Application Load Balancer offers one anycast IP address, routes each user to the nearest healthy backend and supports URL maps, Cloud CDN and Cloud Armor.
Passthrough Network Load Balancers are regional, forward TCP or UDP packets without terminating them and preserve the original client IP address for the backend.
Internal load balancers expose a private IP address inside the VPC, which suits middle tiers that must never be reachable from the internet.
Premium Tier keeps traffic on Google's network to the edge nearest the user, while Standard Tier is cheaper, regional and cannot back a global load balancer.
HA VPN with Cloud Router gives encrypted tunnels over the internet with dynamic routes, whereas Cloud Interconnect gives private dedicated capacity without internet transit.
Choosing storage and databases: Cloud Storage classes, Cloud SQL, Spanner, Bigtable, Firestore and BigQuery
Cloud Storage classes differ in storage price, retrieval fee and minimum storage duration, but every class serves objects with the same low latency when requested.
Cloud SQL is the default for existing MySQL, PostgreSQL or SQL Server applications that fit one region and need managed backups, patching and failover.
Spanner suits relational workloads needing horizontal write scaling and strong consistency across regions, which a single Cloud SQL primary instance cannot provide.
Bigtable stores huge volumes of key-indexed rows with low latency, so it fits IoT and time-series data but offers no joins or multi-row SQL transactions.
BigQuery is built for analytical queries over large datasets and bills by bytes scanned or reserved capacity, so it is not a transactional application database.
Deploying Compute Engine: instance templates, managed instance groups, autoscaling and rolling updates
An instance template is immutable, so changing the machine type or image means creating a new template and rolling the managed instance group onto it.
A regional managed instance group distributes identical VMs across zones, which keeps the application serving when a single zone becomes unavailable.
Autohealing recreates a VM when its application-level health check fails, which catches a hung process that a running VM status would hide.
Autoscaling policies react to CPU utilisation, load balancing capacity, custom metrics or schedules, and always stay within the configured minimum and maximum replicas.
A rolling update replaces instances in batches governed by max surge and max unavailable, and a canary targets only a fraction of the group.
Google Cloud reports Google Associate Cloud Engineer results as pass or fail; it does not publish a numeric passing score.
Which domains does ACE cover?
The official exam guide lists 5 domains: Setting up a cloud solution environment (20%); Planning and configuring a cloud solution (17%); Deploying and implementing a cloud solution (25%); Ensuring successful operation of a cloud solution (20%); Configuring access and security (18%).
Is there a free ACE practice test?
Yes. Gyan AI's free diagnostic is 10 ACE questions with no sign-up, each with an explanation for every option and a link to the official documentation. A free account unlocks the full question set and study notes.
Are these the real ACE exam questions?
No. Real exam content is confidential under vendor exam agreements, and using leaked "dumps" can get a certification revoked. Gyan AI questions are original and written to the published skills outline.