100% Latest & Updated Google Professional Cloud Architect Practice Test Questions, Exam Dumps & Verified Answers!
30 Days Free Updates, Instant Download!
Question 140
A research batch job is restartable and budget-sensitive but can tolerate interruption during excess capacity reclamation. Which operational action best maintains ongoing spot capacity balance after launch?
- Monitor interruption rates and mix standard capacity when completion deadlines tighten
- Assume Spot capacity is always available at the same rate
- Retry every task immediately without backoff during capacity shortage
- Ignore checkpoint write errors to lower cost
Explanation
Correct Answer
A:A cost saving is irrelevant if interruption risk breaches required turnaround
Incorrect Answers
B:Can violate the job deadline. A cost saving is irrelevant if interruption risk breaches required turnaround.
C:May create thrashing. A cost saving is irrelevant if interruption risk breaches required turnaround.
D:Can lose completed work. A cost saving is irrelevant if interruption risk breaches required turnaround.
Question 141
A product team deploys many Kubernetes services but wants managed node operations and consistent workload isolation. Which architecture most directly meets the need for managed Kubernetes infrastructure?
- Choose a manually maintained large fixed GKE Standard cluster for minimal node administration
- Use Cloud Batch to host long-lived customer-facing Kubernetes services
- Select GKE Autopilot if workload constraints fit its managed operating model
- Run each container on separate manually patched VMs
Explanation
Correct Answer
C:Autopilot reduces manual node administration for suitable containerized applications
Incorrect Answers
A:Adds node upkeep unnecessarily. Autopilot reduces manual node administration for suitable containerized applications.
B:Batch runtime does not meet serving orchestration needs. Autopilot reduces manual node administration for suitable containerized applications.
D:Raises node operations workload. Autopilot reduces manual node administration for suitable containerized applications.
Question 142
A product team deploys many Kubernetes services but wants managed node operations and consistent workload isolation. Which implementation choice most directly delivers secure workload identity and scheduling?
- Use the Kubernetes default service identity for every workload with broad cloud roles
- Omit resource requests on all pods
- Set appropriate Kubernetes requests and Workload Identity Federation for GKE bindings
- Store sensitive app credentials in container images
Explanation
Correct Answer
C:Resource requests and workload identity limit privilege and prevent deployment instability
Incorrect Answers
A:Weakens isolation. Resource requests and workload identity limit privilege and prevent deployment instability.
B:Can cause unpredictable scheduling. Resource requests and workload identity limit privilege and prevent deployment instability.
D:Creates persistent secret exposure. Resource requests and workload identity limit privilege and prevent deployment instability.
Question 143
A product team deploys many Kubernetes services but wants managed node operations and consistent workload isolation. Which validation would best establish proof of correct service rollout?
- Check only that Kubernetes accepts YAML manifests
- Test cluster-admin access instead of pod service identity
- Inspect scheduled pod count but not actual requests
- Test readiness, horizontal scaling, and least-privilege cloud API access
Explanation
Correct Answer
D:Pods marked running may still be unable to serve traffic or may hold excessive permissions
Incorrect Answers
A:Not readiness or cloud identity behavior. Pods marked running may still be unable to serve traffic or may hold excessive permissions.
B:Conceals excessive workload permissions. Pods marked running may still be unable to serve traffic or may hold excessive permissions.
C:Does not prove service quality. Pods marked running may still be unable to serve traffic or may hold excessive permissions.
Question 144
A product team deploys many Kubernetes services but wants managed node operations and consistent workload isolation. Which operational action best maintains ongoing cluster efficiency after launch?
- Review workload sizing, deployment availability, and policy violations in GKE
- Assume managed Kubernetes removes need for app security review
- Increase CPU requests uniformly without demand metrics
- Ignore policy denials as long as pods run
Explanation
Correct Answer
A:A managed control plane does not remove application configuration and workload security duties
Incorrect Answers
B:Workload permissions remain customer responsibility. A managed control plane does not remove application configuration and workload security duties.
C:Wastes resources. A managed control plane does not remove application configuration and workload security duties.
D:May hide misconfiguration or exposure. A managed control plane does not remove application configuration and workload security duties.
Question 145
An event-driven order notifier should run only when new Pub/Sub messages arrive and has no long-running sessions. Which architecture most directly meets the need for event-triggered compute?
- Provision a permanent fixed Compute Engine service for a short sporadic handler
- Use Cloud Run functions or a Cloud Run service triggered by supported event delivery
- Use a daily batch job to handle notifications only after 24 hours
- Put all notifications into a user-session cache only
Explanation
Correct Answer
B:Event-driven runtimes avoid a permanently managed VM for short-lived handlers
Incorrect Answers
A:Unnecessary idle infrastructure. Event-driven runtimes avoid a permanently managed VM for short-lived handlers.
C:Does not meet event responsiveness. Event-driven runtimes avoid a permanently managed VM for short-lived handlers.
D:Cannot guarantee asynchronous delivery. Event-driven runtimes avoid a permanently managed VM for short-lived handlers.
Question 146
An event-driven order notifier should run only when new Pub/Sub messages arrive and has no long-running sessions. Which implementation choice most directly delivers retry-safe event handlers?
- Make the handler idempotent and handle at-least-once delivery semantics
- Create side effects before checking whether the event was previously processed
- Acknowledge events before durable downstream completion
- Disable retry and dead-letter features for simplicity
Explanation
Correct Answer
A:Event infrastructure can redeliver a message after failures or lost acknowledgments
Incorrect Answers
B:Can duplicate notifications. Event infrastructure can redeliver a message after failures or lost acknowledgments.
C:Failure can lose work. Event infrastructure can redeliver a message after failures or lost acknowledgments.
D:Transient errors become dropped notifications. Event infrastructure can redeliver a message after failures or lost acknowledgments.
Question 147
An event-driven order notifier should run only when new Pub/Sub messages arrive and has no long-running sessions. Which validation would best establish correctness during redelivery?
- Run a single unique message once
- Check only subscription existence
- Inject the same order event twice and verify only one business notification occurs
- Count upstream event publishes without reading subscriber outputs
Explanation
Correct Answer
C:An HTTP 200 acknowledgement cannot prove downstream idempotency
Incorrect Answers
A:Not enough to prove redelivery safety. An HTTP 200 acknowledgement cannot prove downstream idempotency.
B:Does not exercise the application. An HTTP 200 acknowledgement cannot prove downstream idempotency.
D:Cannot prove one notification per order. An HTTP 200 acknowledgement cannot prove downstream idempotency.
Question 148
An event-driven order notifier should run only when new Pub/Sub messages arrive and has no long-running sessions. Which operational action best maintains operational event completeness after launch?
- Ignore dead-letter routing because Cloud Run is managed
- Track only frontend user sign-ins
- Monitor delivery failures and dead-letter routing for missed notifications
- Delete failed messages immediately without trace
Explanation
Correct Answer
C:Invisible retries can accumulate and conceal dropped business events
Incorrect Answers
A:Application failures remain possible. Invisible retries can accumulate and conceal dropped business events.
B:Does not reveal notifications lost downstream. Invisible retries can accumulate and conceal dropped business events.
D:Prevents repair and replay. Invisible retries can accumulate and conceal dropped business events.
Question 149
A regulated manufacturer must host an unchanged virtualized ERP environment while transitioning its broader infrastructure. Which architecture most directly meets the need for compatibility-first VMware compute?
- Use Google Cloud VMware Engine for an interim VMware-compatible hosting target
- Rewrite the ERP into Cloud Run functions as a precondition of any migration
- Choose a completely different guest OS without application testing
- Use a plain Cloud Storage bucket as the VMware hosting runtime
Explanation
Correct Answer
A:Preserving familiar VMware operations can reduce application rewrites during initial relocation
Incorrect Answers
B:Contradicts short-term compatibility need. Preserving familiar VMware operations can reduce application rewrites during initial relocation.
C:Adds migration risk. Preserving familiar VMware operations can reduce application rewrites during initial relocation.
D:Cannot execute virtual machines. Preserving familiar VMware operations can reduce application rewrites during initial relocation.
Question 150
A regulated manufacturer must host an unchanged virtualized ERP environment while transitioning its broader infrastructure. Which implementation choice most directly delivers private ERP network integration?
- Plan IP address ranges, private connectivity, and directory dependencies before migration
- Migrate VMs before planning ERP subnets and directory access
- Open all ERP VM management ports publicly to simplify support
- Assume all on-prem IP prefixes can be reused unchanged in overlapping VPCs
Explanation
Correct Answer
A:Hypervisor compatibility does not resolve identity and routing dependencies
Incorrect Answers
B:Breaks dependencies. Hypervisor compatibility does not resolve identity and routing dependencies.
C:Creates unacceptable exposure. Hypervisor compatibility does not resolve identity and routing dependencies.
D:Causes routing conflicts. Hypervisor compatibility does not resolve identity and routing dependencies.
Google Cloud Certified - Professional Cloud Architect (Professional Cloud Architect) Practice Tests