GCP-PCA Practice Questions — Free

Question 1 of 25

A startup needs a relational database for a new product with unknown scale and wants to avoid a migration later if it succeeds globally. What is the pragmatic choice?

• A.Spanner multi-region from day one regardless of the current budget
• B.Cloud Storage objects used as the primary transactional database
• C.Bigtable configured to behave as a relational database
• D.Cloud SQL now, migrating to Spanner only if global scale materializesCorrect

Question 2 of 25

KnightMotives wants a CRM to track customer interactions and personalize experiences. Which disposition is most defensible?

• A.Track customer interactions in shared spreadsheets
• B.Buy a SaaS CRM and integrate via APIs and BigQueryCorrect
• C.Use the ERP mainframe as the customer relationship system
• D.Build a custom CRM on GKE from scratch in-house

Question 3 of 25

A retailer's flash-sale traffic is 20x normal for two hours a month. Which compute choice best matches cost and elasticity on Google Cloud?

• A.A single large Compute Engine VM sized for the flash-sale peak
• B.Reserved Compute Engine capacity sized for the peak and held year-round
• C.Sole-tenant nodes provisioned for the peak and idle otherwise
• D.Cloud Run or GKE Autopilot scaling on demand from a small steady baselineCorrect

Question 4 of 25

A company is designing for a workload that reads the same reference dataset millions of times per second with rare updates. What layer should be added?

• A.A larger Cloud SQL instance provisioned to absorb the reads
• B.Memorystore or in-process caching with invalidation on updateCorrect
• C.Reading the dataset directly from Cloud Storage on each request
• D.BigQuery queries issued for every individual read request

Question 5 of 25

KnightMotives targets autonomous features first in regions with favorable regulation. What architectural implication follows?

• A.Storing all vehicle telemetry in one US region for simplicity
• B.Deploying autonomous features everywhere and disabling them later
• C.Region-scoped deployments and residency controls enabled per jurisdictionCorrect
• D.A single global deployment that treats every region identically

Question 6 of 25

A company's non-functional requirements include 200 ms p95 latency worldwide for an API. Which design element is most directly responsible for meeting it?

• A.Multi-region deployment with global anycast routing and data localityCorrect
• B.A larger request timeout configured on the load balancer
• C.A single region deployment with a larger machine type
• D.Weekly performance reviews by the architecture team

Question 7 of 25 · multiple response

Which of the following are valid options for running scheduled jobs on Google Cloud without a persistent VM? (Select TWO)

• A.Cloud Scheduler triggering a Cloud Run jobCorrect
• B.Cron on a permanently running Compute Engine VM
• C.A developer running the job manually each day
• D.Cloud Scheduler publishing to Pub/Sub consumed by a functionCorrect

Question 8 of 25

A team must provision a private GKE cluster whose nodes have no public IPs but must still pull images and reach Google APIs. What is required?

• A.Public IP addresses on every node with a restrictive firewall
• B.No networking changes, since private nodes reach the internet
• C.A Cloud VPN tunnel to Google's API endpoints from the cluster
• D.Private Google Access on the subnet plus Cloud NAT for other egressCorrect

Question 9 of 25

Altostrat must provision scalable Kubernetes on-premises for ingestion while cloud clusters handle delivery. Which fleet feature keeps configuration consistent across both?

• A.Manual kubectl apply performed on each cluster by operators
• B.Separate config repositories per cluster with no synchronization
• C.Config Sync and Policy Controller across all fleet clustersCorrect
• D.Copying YAML manifests between clusters by email attachment

Question 10 of 25 · multiple response

EHR Healthcare needs Redis and MongoDB equivalents managed on Google Cloud. Which mappings fit? (Select TWO)

• A.Memorystore for RedisCorrect
• B.Cloud CDN for MongoDB
• C.BigQuery for Redis caching
• D.MongoDB Atlas on Google Cloud or Firestore depending on data modelCorrect

Question 11 of 25

A company must comply with PCI DSS while using Google Cloud. Which statement is correct about responsibility?

• A.PCI DSS requirements do not apply to workloads running in the cloud
• B.Compliance responsibility transfers entirely to the payment processor
• C.Google provides compliant infrastructure; the customer owns configurationCorrect
• D.Google handles all PCI compliance obligations on behalf of its customers

Question 12 of 25

A SaaS provider must isolate each tenant's data cryptographically. Which pattern fits?

• A.Encryption keys stored in the application's own database tables
• B.One encryption key shared across every tenant's data
• C.No encryption at rest applied to any of the tenant data
• D.Per-tenant customer-managed keys, with EKM for premium tiersCorrect

Question 13 of 25

A company must retain audit logs for seven years for compliance and prevent tampering. Which configuration fits?

• A.Audit log entries copied into a spreadsheet every quarter
• B.A locked-retention bucket sink, or a log bucket with locked retentionCorrect
• C.Deleting audit logs quarterly after a manual compliance review
• D.The default 30-day retention in the _Default log bucket

Question 14 of 25

An architect must design secrets management for many services. Which approach fits?

• A.Secrets hard-coded inside the container images at build time
• B.Secret Manager with per-service IAM, rotation, and audit loggingCorrect
• C.Secrets in environment files committed to the Git repository
• D.A shared spreadsheet listing every service's credentials

Question 15 of 25

An organization wants to enforce that all new buckets use CMEK and uniform access. What is the correct control?

• A.IAM Viewer roles granted broadly across the organization
• B.Organization policy constraints requiring CMEK and uniform bucket accessCorrect
• C.A wiki page describing the required bucket settings for all teams
• D.Periodic manual audits of existing buckets by the security team

Question 16 of 25

KnightMotives' financial priorities must shift globally to fund AI. Which analysis supports the decision?

• A.A single vendor quote covering the whole modernization program
• B.A multi-year TCO and ROI model comparing phased modernization optionsCorrect
• C.Deferring all financial analysis to the CFO's office without input
• D.Estimates derived from competitor press releases and analyst notes

Question 17 of 25

A company wants developers to self-provision approved infrastructure without ticket queues. Which approach fits?

• A.Manual provisioning by a single administrator on request
• B.Granting the Owner role on the organization to all developers
• C.A service catalog of vetted Terraform modules through a platform portalCorrect
• D.A monthly provisioning window handled by the platform team

Question 18 of 25 · multiple response

Which of the following are stakeholder-management practices during a cloud program? (Select TWO)

• A.Withholding status from executives
• B.Changing scope without communication
• C.Clear escalation paths for cross-team conflictsCorrect
• D.Regular steering reviews with decision logsCorrect

Question 19 of 25

EHR Healthcare needs a service catalog so hospital onboarding is faster. What should the catalog contain?

• A.Approved Terraform modules for tenant environments with policy checksCorrect
• B.A folder of screenshots documenting the console steps
• C.Ad hoc provisioning scripts maintained by each engineer
• D.A list of employee phone numbers organized by department

Question 20 of 25 · multiple response

Which of the following are good practices when handing a delivered platform to an operations team? (Select TWO)

• A.No documentation, since the code is self-explanatory
• B.Handover only after the first major outage
• C.Runbooks for common failures and escalation pathsCorrect
• D.Dashboards and alerts already configured against SLOsCorrect

Question 21 of 25

Cymbal must expose catalog and order APIs to a mobile app and third-party marketplaces with quotas and analytics. Which service fits the guide's API management item?

• A.Direct database access through a public IP with a firewall rule
• B.Apigee with API products, a developer portal, quotas, and analyticsCorrect
• C.A load balancer in front of the services with no authentication
• D.An SFTP endpoint where partners upload and download JSON files

Question 22 of 25

A team implementing a new platform has no automated tests for infrastructure. What should the architect require?

• A.Manual peer review of each change with no automated checks at all
• B.Skipping tests entirely in order to meet the delivery deadline
• C.Policy-as-code checks and infrastructure tests in CI before any shared applyCorrect
• D.Testing infrastructure changes directly in the production environment

Question 23 of 25

Altostrat currently gets alerts only by email from mixed Cloud Monitoring and Prometheus setups. Which change best improves operational excellence?

• A.Adding more email recipients to the existing Prometheus alerts
• B.Disabling Prometheus and relying on Cloud Monitoring defaults
• C.Manual dashboard checks every hour by the operations team
• D.Managed Prometheus into Cloud Monitoring, alerts routed to on-call channelsCorrect

Question 24 of 25

A team wants to ensure reliability under load before a launch. Which practice fits section 6.6?

• A.Launching the service and watching how it behaves under real traffic
• B.Unit tests only, since they exercise every code path
• C.Load testing beyond expected peak with observed SLOs and breaking pointsCorrect
• D.Vendor assurances that the platform scales automatically

Question 25 of 25 · multiple response

Which of the following are recommended observability practices for hybrid workloads under the operational excellence pillar? (Select TWO)

• A.Sampling all logs to zero
• B.Separate dashboards nobody correlates
• C.Unified logging and metrics across on-prem and cloud in Cloud OperationsCorrect
• D.Distributed tracing across service boundariesCorrect