Google Professional Cloud Architect (PCA) Practice Test
Design, Secure, and Optimize Google Cloud Architectures.
500 scenario-based practice questions covering all 6 PCA exam sections: solution design, infrastructure provisioning, security and compliance, process optimization, implementation, and operations excellence. Mapped to the official exam guide.
No credit card required · Upgrade to 500 questions for $11.99
500 Exam-Style Questions
7-Day Money-Back Guarantee
Is this for you?
You design on Google Cloud already but want practice choosing between GKE, Cloud Run, and Cloud Run functions when a scenario gives cost, latency, and scaling limits together
You are preparing for questions tied to the Altostrat Media, Cymbal Retail, EHR Healthcare, and KnightMotives Automotive case studies and want to practice pulling requirements out of a long brief
You know IAM basics but want drills on resource hierarchy, separation of duties, Cloud KMS customer-managed keys, and VPC Service Controls in the same design
You need to plan migrations with Google Cloud Migration Center and weigh license implications, dependency planning, and workload disposition (build, buy, modify, or deprecate)
Not for you if:
You are new to Google Cloud and still learning how to create projects, VMs, and networks: the Associate Cloud Engineer exam is the better first step
You want recalled or leaked exam questions: CertSharp does not offer braindumps, only original questions written against the PCA exam guide
The Google Professional Cloud Architect (PCA) is a professional-level Google Cloud certification. The exam has 50–60 questions, runs 2 hours, costs $200, and requires a passing score of Not published by Google. CertSharp provides 500 exam-style PCA practice questions — 30 free with no credit card — each with a detailed explanation.
Designing and planning a cloud solution architecture (~25%)
2
Managing and provisioning a cloud solution infrastructure (~17.5%)
3
Designing for security and compliance (~17.5%)
4
Analyzing and optimizing technical and business processes (~15%)
5
Managing implementation (~12.5%)
6
Ensuring solution and operations excellence (~12.5%)
From the exam guide
What the PCA exam actually tests
The guide treats the Well-Architected Framework pillars (operational excellence, security, reliability, performance optimization, cost optimization, and sustainability) as a thread through every section, so most questions ask for the design that best fits stated business and technical requirements. Several of the case study businesses use Google Cloud generative AI, so AI design choices appear inside architecture scenarios, not only as a separate topic.
Business requirements before products
Section 1.1 lists KPIs, ROI, business continuity, integration with external systems, and workload disposition. You must read a scenario, separate functional from non-functional requirements, and pick the design whose trade-offs match what the business said it values.
Network and compute design
Expect hybrid and multicloud connectivity, Shared VPC, peering, hierarchical firewall policy, load balancer choice, and Private Service Connect. On compute, you decide between spot and standard VMs, custom machine types, GKE, Cloud Run, Cloud Run functions, and Google Cloud VMware Engine.
AI workloads in the architecture
Sections 2.4 and 2.5 cover Agent Platform Pipelines, GPUs and TPUs through AI Hypercomputer, Model Garden models, and choosing among Google AI APIs for search, conversation, vision, video, and audio. Section 3.1 adds securing AI with Model Armor and Sensitive Data Protection.
Compliance and operations
Compliance questions cover health record privacy, children's privacy, data sovereignty, card data and PII, and SOC 2 audits backed by logs. Operations questions cover alerting strategy, profiling, release management, and reliability testing such as chaos engineering, penetration testing, and load testing.
Why CertSharp
Built for Exam Day, Not Just Study Time
Study guides teach concepts. We prepare you for the actual PCA exam.
1
Architecture trade-off questions: pick the best design for stated business and technical requirements, not just a service that works
2
Networking and security scenarios go deep, including Private Service Connect, Shared VPC, VPC Service Controls, and IAM design
3
Covers every section of the official exam guide, including migration planning, compliance, and operational excellence
4
Aligned to the current exam guide, including Gemini Enterprise Agent Platform, Model Garden, and AI Hypercomputer topics
Free Sample Questions
PCA Practice Questions (Free Sample)
Real PCA exam-style questions from our 500-question Google Professional Cloud Architect bank, each with the correct answer and a detailed explanation.
Question 1 — PCA
Pods on a GKE cluster must call Cloud Storage and BigQuery as a dedicated Google identity, and the security team prohibits mounting service account key files into pods. What is the recommended approach?
A
Grant the node pool's default Compute Engine service account the Owner role on the project
B
Embed a user's OAuth refresh token in the container image as an environment variable
C
Use Workload Identity Federation for GKE so the Kubernetes service account maps to an IAM principal
D
Store a service account key in a Kubernetes Secret and mount it at GOOGLE_APPLICATION_CREDENTIALS
Correct answer: C — Explanation
Workload Identity Federation for GKE lets pods obtain short-lived credentials through their Kubernetes service account, either via IAM principal bindings or impersonating a Google service account, with no keys. A Secret-mounted key is still a long-lived key. Giving the node service account Owner breaks least privilege for every pod on the node. A refresh token ties production to a person's credential and leaks through image layers.
Question 2Managing and provisioning a cloud solution infrastructureEasy
Knowledge workers want an AI research assistant that summarizes only the internal documents they upload and cites passages, without writing any code. Which product fits?
A
Gemini Notebook Enterprise (formerly NotebookLM Enterprise), with the team's uploaded documents as its sources
B
BigQuery ML text generation running over a table of the document file names
C
Cloud Natural Language API classification of each document into predefined categories
D
A custom retrieval pipeline on GKE with an open-weight model and a vector database
Show answer & explanation
Correct answer: A — Gemini Notebook Enterprise (formerly NotebookLM Enterprise), with the team's uploaded documents as its sources
Gemini Notebook Enterprise (formerly NotebookLM Enterprise) grounds its responses in the sources users upload and needs no development. Building a custom retrieval stack is far more work for the same need. Generating text over file names has no access to document content. Classification labels documents but does not answer questions.
Question 3Analyzing and optimizing technical and business processesEasy
Which Cloud Build capability starts a build automatically when a developer pushes a commit to a specific branch of a connected repository?
A
A build trigger configured with the repository, a branch regex, and a build config file
B
A Cloud Deploy delivery pipeline that compiles source code when a release is created
C
A Pub/Sub subscription attached directly to an Artifact Registry repository for pushes
D
A Cloud Scheduler job that polls the repository and calls the Cloud Build list builds API
Show answer & explanation
Correct answer: A — A build trigger configured with the repository, a branch regex, and a build config file
Build triggers connect a repository event, such as a push to a branch matching a regex, to a build configuration. Polling with Cloud Scheduler is unnecessary and does not start builds from source events. Artifact Registry publishes events after images are pushed, not when source is committed. Cloud Deploy handles delivery of already-built artifacts and does not compile code.
Question 4Designing for security and complianceEasy
A team wants container images pushed to Artifact Registry to be automatically checked for known vulnerabilities in operating system packages. Which feature provides this?
A
Cloud CDN caching of image layers close to the cluster
B
Cloud Scheduler jobs that pull each image and run a shell script
C
Cloud NAT logging for outbound requests from the build workers
D
Artifact Analysis vulnerability scanning enabled for the repository
Show answer & explanation
Correct answer: D — Artifact Analysis vulnerability scanning enabled for the repository
Artifact Analysis scans images stored in Artifact Registry for known vulnerabilities and records the results as occurrences. Cloud Scheduler scripts would be self-built and unmanaged. Cloud CDN caches content and Cloud NAT logging records network translations, so neither identifies vulnerable packages.
Question 5Ensuring solution and operations excellenceEasy
A newly deployed Cloud Run revision is raising errors. What is the fastest way to restore service?
A
Route 100% of traffic back to the previous healthy revision
B
Increase the maximum instance count for the service to the highest value
C
Switch the service to a different region and update the DNS records manually
D
Delete the Cloud Run service and recreate it from the source repository
Show answer & explanation
Correct answer: A — Route 100% of traffic back to the previous healthy revision
Cloud Run retains previous revisions, so shifting traffic back to the last good revision is immediate and requires no rebuild. Deleting the service causes downtime and loses configuration. More instances multiplies the faulty code, and moving regions is slow and unrelated to the bad revision.
Question 6Designing and planning a cloud solution architectureEasy
A mobile app team needs a serverless NoSQL document database with real-time synchronization to clients and automatic scaling, without managing servers. Which service fits best?
A
Cloud SQL for PostgreSQL, a managed relational database with fixed instance sizing
B
Firestore, a serverless document database with real-time listeners and offline support
C
Bigtable, a wide-column store sized by the number of provisioned nodes in a cluster
D
Memorystore for Redis, an in-memory cache that holds sessions on provisioned instances
Show answer & explanation
Correct answer: B — Firestore, a serverless document database with real-time listeners and offline support
Firestore is a serverless document database with real-time listeners, offline support, and automatic scaling, suited to mobile and web back ends. Cloud SQL is relational and sized per instance. Bigtable targets large analytical and time-series workloads and needs node provisioning. Memorystore is a cache and not a durable primary database.
Question 7Managing implementationMedium
A marketing team wants daily loads of Google Ads and YouTube channel reports into BigQuery without writing ETL code. Which service should they configure?
A
Datastream with a Google Ads source and BigQuery destination
B
BigQuery Data Transfer Service with scheduled transfers from the supported connectors
C
Dataflow streaming jobs written to poll each product API once a minute
D
Storage Transfer Service jobs that copy report exports into Cloud Storage manually
Show answer & explanation
Correct answer: B — BigQuery Data Transfer Service with scheduled transfers from the supported connectors
BigQuery Data Transfer Service automates scheduled, managed loads from supported sources such as Google Ads and YouTube reports directly into BigQuery with no code. Datastream captures database changes and has no Google Ads connector. Custom Dataflow polling is code the team would have to build and run. Storage Transfer Service moves objects, not API-based report data.
Question 8Managing and provisioning a cloud solution infrastructureMedium
A Cloud Run service must connect to a Cloud SQL instance using its private IP in a VPC. The team prefers the option that avoids managing connector VMs. What should they configure?
A
An ingress setting of internal so that Cloud SQL can call back into the service
B
Direct VPC egress on the service, specifying the network and subnet
C
A Cloud NAT gateway on the subnet, which gives Cloud Run a route to the private IP
D
A Serverless VPC Access connector with a /28 range and instances sized for peak traffic
Show answer & explanation
Correct answer: B — Direct VPC egress on the service, specifying the network and subnet
Direct VPC egress lets Cloud Run send traffic into a VPC subnet without connector instances. A Serverless VPC Access connector also works but requires provisioning and sizing connector capacity, which the team wants to avoid. Ingress settings control inbound requests only. Cloud NAT does not give Cloud Run a VPC path by itself.
Question 9Analyzing and optimizing technical and business processesMedium
Engineering teams view cloud cost as the finance department's problem, and budgets are regularly exceeded. Which FinOps practice best addresses this?
A
Shared accountability, with cost visibility per team and regular reviews among engineering, finance, and product
B
Move the cost decisions to the finance department to centralize control over cloud spending
C
Remove the cost dashboards from engineers so that they can focus on delivering new features
D
Cap each team's budget at last year's spend, regardless of how the business has grown
Show answer & explanation
Correct answer: A — Shared accountability, with cost visibility per team and regular reviews among engineering, finance, and product
FinOps is a cultural practice in which engineering, finance, and product share accountability, supported by visibility, allocation, and recurring optimization cycles. Centralizing decisions in finance leaves out those who control usage. Fixed caps ignore growth. Hiding cost data prevents informed trade-offs.
Question 10Designing for security and complianceMedium
Developers need to run Terraform with the permissions of a privileged deployment service account, but the security team does not want any service account keys created. What is the recommended approach?
A
Add the developers as members of the service account's default Compute Engine group
B
Grant roles/iam.serviceAccountTokenCreator on the service account to the developers' group and have them impersonate it
C
Grant the developers roles/owner on the project so they do not need the service account
D
Create a key for the service account and distribute it through a shared password manager
Show answer & explanation
Correct answer: B — Grant roles/iam.serviceAccountTokenCreator on the service account to the developers' group and have them impersonate it
Service account impersonation lets authorized users mint short-lived tokens for the service account, with audit logs tying the action to the real user, and no key exists. Owner is excessive and bypasses the intended permission boundary. Distributing a key creates the long-lived credential the team wants to avoid. Compute Engine has no such default group.
Want more PCA practice questions?
Take the free PCA practice test: 30 questions per session with instant explanations. Unlock all 500 questions plus timed exam mode for $11.99 one-time.
Everything you need to know about Google Professional Cloud Architect
What is the Google Professional Cloud Architect certification?
The Google Professional Cloud Architect (PCA) validates your ability to design, develop, and manage secure, scalable, highly available, and cost-effective solutions on Google Cloud. Cost: $200, 50-60 questions, 2-hour time limit.
How hard is the Google PCA exam?
PCA is a professional-level exam and one of the more demanding Google Cloud certifications. It tests architectural judgment: choosing among several plausible designs based on business requirements, cost, security, and reliability. Case study questions make up 20-30% of the exam, so reading requirements carefully matters as much as product knowledge.
Is Google PCA worth it in 2026?
For architects, cloud consultants, and senior engineers working on Google Cloud, PCA is one of the most recognized credentials on the platform. It demonstrates design and decision-making skill rather than only hands-on configuration. Whether it is worth it depends on your role and how much Google Cloud work you do.
What experience does Google recommend for PCA?
Google lists no prerequisites. The recommended experience is 3+ years of industry experience including 1+ years designing and managing solutions using Google Cloud. Associate Cloud Engineer is not required first.
What topics are on the Google PCA exam?
The six exam sections are solution architecture design and planning, infrastructure provisioning, security and compliance, technical and business process optimization, managing implementation, and operations excellence. Topics include VPC and hybrid networking, storage and compute selection, IAM and resource hierarchy, VPC Service Controls, migration planning, CI/CD, observability, disaster recovery, and Google Cloud AI solutions.
How many questions are on the Google PCA exam?
The standard PCA exam has 50-60 multiple choice and multiple select questions with a 2-hour time limit. Each exam includes 2 case studies. Google does not publish a passing score.
How long to prepare for Google PCA?
It varies with your background. Candidates with solid Google Cloud design experience often spend several weeks, while those newer to the platform may need longer. Work through the official learning path, study the four published case studies, and practice reading scenarios for the real requirement.
What is the difference between Google PCA and Associate Cloud Engineer?
ACE validates associate-level implementation skills: deploying and managing Google Cloud solutions. PCA validates professional-level design skills: choosing architectures that meet business and technical requirements. PCA questions emphasize trade-offs and case studies rather than individual commands.
How much does the Google Professional Cloud Architect exam cost?
The standard Google PCA exam costs $200 USD plus applicable tax. It runs 2 hours and can be taken at a testing center or online with remote proctoring. Google also offers a shorter renewal exam for existing certificate holders.
Is there a free Google PCA practice test?
Yes. CertSharp offers a free Google PCA practice test: 30 questions per session with detailed explanations, no credit card required. The full 500-question Google PCA bank, including timed exam mode (60 questions), is $11.99 one-time with lifetime access.
Also on CertSharp
Related to the PCA Exam
Where PCA candidates often go next, each with its own 500-question bank.