Google Professional Cloud Architect Practice Test 91
Question 900
Consider National small-business bank and its finops reserved capacity workload. The issue is that leadership commits all compute to three-year reservations despite demand reduction plans; the current observation is that forecasts show the main batch system will retire within six months. The team must ensure that reserved spend must not exceed expected future compute usage. Which continuous monitoring approach best reveals recurrence or changing dependencies?
- Monitor commitment utilization and forecast changes
- Buy maximal three-year reservations for workloads scheduled for decommission
- Compare billed baseline to actual consumption through retirement
- Choose commitments only for stable baseline needs and leave uncertain capacity flexible
Explanation
Correct Answer
A:
Monitor commitment utilization and forecast changes reveals recurrence, drift or changing dependencies after launch. It provides a repeatable signal for the risk (leadership commits all compute to three-year reservations despite demand reduction plans) rather than treating the first successful test as permanent assurance.
Incorrect Answers
B:
Long obligations can outlast the resource demand. For this question, it is not a defensible choice for ongoing operational assurance and it does not satisfy the requirement that reserved spend must not exceed expected future compute usage.
C:
The alternative “Compare billed baseline to actual consumption through retirement” is a one-time acceptance result rather than ongoing operational assurance. In the reported observation, forecasts show the main batch system will retire within six months. The option “Monitor commitment utilization and forecast changes” instead addresses the requested ongoing operational assurance.
D:
The alternative “Choose commitments only for stable baseline needs and leave uncertain capacity flexible” selects a design, but does not provide the ongoing signal required to detect drift. In the reported observation, forecasts show the main batch system will retire within six months. The option “Monitor commitment utilization and forecast changes” instead addresses the requested ongoing operational assurance.
Question 901
At Digital university archive, the BigQuery scripted exports workload has a documented risk: a nightly bq CLI export uses wildcard datasets and leaks unrelated faculty research. The review found that audit log shows query scope expanded after a new dataset was added. The architecture must also ensure that exports must include only approved course statistics. What follow-up metric or alert will identify emerging failures after the initial acceptance test?
- Track permission changes and unexpected query bytes
- Create an unrelated dataset and verify the export excludes it
- Use parameterized explicit BigQuery dataset/table scope with read-only job identity
- Give the export service account BigQuery Admin over the whole organization
Explanation
Correct Answer
A:
Track permission changes and unexpected query bytes reveals recurrence, drift or changing dependencies after launch. It provides a repeatable signal for the risk (a nightly bq CLI export uses wildcard datasets and leaks unrelated faculty research) rather than treating the first successful test as permanent assurance.
Incorrect Answers
B:
The alternative “Create an unrelated dataset and verify the export excludes it” is a one-time acceptance result rather than ongoing operational assurance. In the reported observation, audit log shows query scope expanded after a new dataset was added. The option “Track permission changes and unexpected query bytes” instead addresses the requested ongoing operational assurance.
C:
The alternative “Use parameterized explicit BigQuery dataset/table scope with read-only job identity” selects a design, but does not provide the ongoing signal required to detect drift. In the reported observation, audit log shows query scope expanded after a new dataset was added. The option “Track permission changes and unexpected query bytes” instead addresses the requested ongoing operational assurance.
D:
Broad admin rights violate the limited read-only extraction purpose. For this question, it is not a defensible choice for ongoing operational assurance and it does not satisfy the requirement that exports must include only approved course statistics.
Question 902
Manufacturer predictive maintenance team plans to update Agent Platform ML pipelines. Investigators found that model results diverge even when training data IDs match, consistent with the reported problem that each retraining run uses a different preprocessing notebook with no lineage. The response must satisfy the requirement that factory anomaly models need reproducible releases and operator approval. Which target-state design is best justified by the evidence and required business outcome?
- Train directly on mutable shared notebooks without tracking artifact versions
- Compare data hashes, preprocessing versions, evaluation steps and feature schema
- Use versioned ML pipelines with recorded artifacts and reproducible evaluation gates
- Monitor pipeline failures, model drift and promotion exceptions
Explanation
Correct Answer
C:
Use versioned ML pipelines with recorded artifacts and reproducible evaluation gates targets the underlying issue (each retraining run uses a different preprocessing notebook with no lineage) rather than merely reporting symptoms. It also preserves the required outcome (factory anomaly models need reproducible releases and operator approval), which a target-state design must satisfy.
Incorrect Answers
A:
An unversioned notebook cannot support reproducible evaluation. For this question, it is not a defensible choice for architecture selection and it does not satisfy the requirement that factory anomaly models need reproducible releases and operator approval.
B:
The alternative “Compare data hashes, preprocessing versions, evaluation steps and feature schema” investigates the cause without specifying the durable architecture to resolve it. In the reported observation, model results diverge even when training data IDs match. The option “Use versioned ML pipelines with recorded artifacts and reproducible evaluation gates” instead addresses the requested architecture selection.
D:
The alternative “Monitor pipeline failures, model drift and promotion exceptions” measures results without defining the preventive design now required. In the reported observation, model results diverge even when training data IDs match. The option “Use versioned ML pipelines with recorded artifacts and reproducible evaluation gates” instead addresses the requested architecture selection.
Question 903
Case study: International cinema platform: secure software supply chain
International cinema platform relies on secure software supply chain to deliver its operational commitments. The current architectural problem is that a deployment references mutable container tags from an unaudited third-party registry. The team has recorded this evidence: build provenance report shows tag digest changed after approval. Any proposed response must preserve the requirement that production media services must run only approved signed artifacts.
The secure software supply chain environment at International cinema platform faces this issue: a deployment references mutable container tags from an unaudited third-party registry. Supporting evidence shows that build provenance report shows tag digest changed after approval. The architect must preserve the requirement that production media services must run only approved signed artifacts. Select THREE evidence activities to cover (1) diagnosis before design, (2) direct post-change acceptance, and (3) continued operational monitoring. Which THREE fit those stages?
- Trace build provenance, registry trust and admission permissions
- Use Artifact Registry provenance and Binary Authorization enforcement for approved images
- Exclude the affected dependency from all future tests and incident metrics to simplify reporting
- Attempt deployment with unsigned and digest-swapped images and verify rejection
- Allow every external image when its tag says latest-stable
- Watch policy bypasses, signer changes and rejected admissions
Explanation
Correct Answers
A:
Trace build provenance, registry trust and admission permissions is the appropriate diagnostic because the observation—build provenance report shows tag digest changed after approval—does not by itself establish every dependency, authorization rule or failure boundary. The findings explain the reported risk (a deployment references mutable container tags from an unaudited third-party registry) before the design is approved. It supplies the pre-design diagnostic evidence requested.
D:
Attempt deployment with unsigned and digest-swapped images and verify rejection is a direct acceptance test of the path implicated by build provenance report shows tag digest changed after approval. It checks that the original risk (a deployment references mutable container tags from an unaudited third-party registry) has actually been controlled and tests that permitted operations still meet the requirement that production media services must run only approved signed artifacts. It supplies the separate post-change acceptance result requested.
F:
Watch policy bypasses, signer changes and rejected admissions reveals recurrence, drift or changing dependencies after launch. It provides a repeatable signal for the risk (a deployment references mutable container tags from an unaudited third-party registry) rather than treating the first successful test as permanent assurance. It supplies the ongoing observation requested.
Incorrect Answers
B:
Designing secure software supply chain addresses the root cause, but this specific three-part question asks for evidence to diagnose, accept and continuously verify the response to a deployment references mutable container tags from an unaudited third-party registry, not another design proposal.
C:
Omitting the relevant dependency from testing and monitoring would hide, not correct, recurrence of the failure. In particular, the observed issue was: build provenance report shows tag digest changed after approval.
E:
A mutable tag does not establish artifact authenticity. It cannot satisfy the required evidence stages.
Question 904
Consider Emergency veterinary service and its 24-hour booking network workload. The issue is that the migration estimate ignores the impact of a failed regional booking queue; the current observation is that incident records show bookings lost when the only queue broker restarts. The team must ensure that emergency hospitals must reconcile bookings after a site outage. The change committee needs to understand the current system rather than install a fix yet. What should it do?
- Calculate booking RPO/RTO, delivery guarantees and regional dependency chains
- Dual-write for a controlled pilot with deterministic booking identifiers
- Build a durable booking pipeline with idempotent consumers and multi-zone service recovery
- Use an in-memory single-instance queue for all accepted bookings
Explanation
Correct Answer
A:
Calculate booking RPO/RTO, delivery guarantees and regional dependency chains is the appropriate diagnostic because the observation—incident records show bookings lost when the only queue broker restarts—does not by itself establish every dependency, authorization rule or failure boundary. The findings explain the reported risk (the migration estimate ignores the impact of a failed regional booking queue) before the design is approved.
Incorrect Answers
B:
The alternative “Dual-write for a controlled pilot with deterministic booking identifiers” is rollout work that may modify the wrong dependency before root-cause evidence is collected. In the reported observation, incident records show bookings lost when the only queue broker restarts. The option “Calculate booking RPO/RTO, delivery guarantees and regional dependency chains” instead addresses the requested root-cause evidence.
C:
The alternative “Build a durable booking pipeline with idempotent consumers and multi-zone service recovery” is target architecture selection; it does not establish why the observed path failed. In the reported observation, incident records show bookings lost when the only queue broker restarts. The option “Calculate booking RPO/RTO, delivery guarantees and regional dependency chains” instead addresses the requested root-cause evidence.
D:
A volatile single broker loses accepted work on restart. For this question, it is not a defensible choice for root-cause evidence and it does not satisfy the requirement that emergency hospitals must reconcile bookings after a site outage.
Question 905
Case study: Public school admissions portal: release version safety
Public school admissions portal relies on release version safety to deliver its operational commitments. The current architectural problem is that frontend rollout references a backend schema that is not yet deployed. The team has recorded this evidence: incident shows new clients send unsupported enrollment fields. Any proposed response must preserve the requirement that admission forms cannot break during deadline week.
Consider Public school admissions portal and its release version safety workload. The issue is that frontend rollout references a backend schema that is not yet deployed; the current observation is that incident shows new clients send unsupported enrollment fields. The team must ensure that admission forms cannot break during deadline week. The change committee needs to understand the current system rather than install a fix yet. What should it do?
- Roll out schema first, then canary clients with traffic rollback
- Design backward-compatible API contracts and staged deployments
- Deploy breaking frontend and database schema simultaneously without canary
- Map frontend/backend compatibility and schema transition dependencies
Explanation
Correct Answer
D:
Map frontend/backend compatibility and schema transition dependencies is the appropriate diagnostic because the observation—incident shows new clients send unsupported enrollment fields—does not by itself establish every dependency, authorization rule or failure boundary. The findings explain the reported risk (frontend rollout references a backend schema that is not yet deployed) before the design is approved.
Incorrect Answers
A:
The alternative “Roll out schema first, then canary clients with traffic rollback” is rollout work that may modify the wrong dependency before root-cause evidence is collected. In the reported observation, incident shows new clients send unsupported enrollment fields. The option “Map frontend/backend compatibility and schema transition dependencies” instead addresses the requested root-cause evidence.
B:
The alternative “Design backward-compatible API contracts and staged deployments” is target architecture selection; it does not establish why the observed path failed. In the reported observation, incident shows new clients send unsupported enrollment fields. The option “Map frontend/backend compatibility and schema transition dependencies” instead addresses the requested root-cause evidence.
C:
A coupled big-bang change creates untested integration failures. For this question, it is not a defensible choice for root-cause evidence and it does not satisfy the requirement that admission forms cannot break during deadline week.
Question 906
At National agriculture ministry, the sovereign data retention workload has a documented risk: analysts want one globally replicated bucket for personal farm ownership records. The review found that compliance map lists a specific country-only data-storage rule. The architecture must also ensure that regulatory data must remain within the designated geography. Select TWO complementary actions: one durable target architecture and one controlled way to introduce it while preserving service. Which TWO are best?
- Select location-restricted storage and explicit organization location policies
- Audit location-policy drift and cross-border egress activity
- Rehome eligible object groups with controlled object-copy checks and deny policies
- Use a multi-region spanning prohibited locations for all identifiers
- Verify exact residency constraints, access jurisdictions and lawful transfers
Explanation
Correct Answers
A:
Select location-restricted storage and explicit organization location policies targets the underlying issue (analysts want one globally replicated bucket for personal farm ownership records) rather than merely reporting symptoms. It also preserves the required outcome (regulatory data must remain within the designated geography), which a target-state design must satisfy. This supplies the lasting-control half of the two-action package.
C:
Rehome eligible object groups with controlled object-copy checks and deny policies is a controlled means to introduce the approved architecture while meeting the requirement that regulatory data must remain within the designated geography. Staged delivery, attributable change evidence and rollback guard against an otherwise avoidable service disruption. This supplies the controlled-introduction half of the two-action package.
Incorrect Answers
B:
Monitoring sovereign data retention for future drift is useful after the rollout, but it does not specify the requested target control and safe implementation for this case: analysts want one globally replicated bucket for personal farm ownership records.
D:
A multi-region location cannot be assumed compliant with a country-only restriction. The shortcut does not supply the requested design and safe rollout for sovereign data retention, where compliance map lists a specific country-only data-storage rule.
E:
This diagnoses the original issue (analysts want one globally replicated bucket for personal farm ownership records), but the TWO requested actions are the target architecture and its implementation; discovery was already completed.
Question 907
At Digital university archive, the Google Cloud design for BigQuery scripted exports is affected because a nightly bq CLI export uses wildcard datasets and leaks unrelated faculty research. A review found that audit log shows query scope expanded after a new dataset was added. The proposed response must account for the requirement that exports must include only approved course statistics. A deployment ticket has been closed. What direct evidence would justify solution signoff?
- Deploy export script with bounded project context and approved schedules
- Identify exact dataset boundaries, IAM roles and query requirements
- Give the export service account BigQuery Admin over the whole organization
- Create an unrelated dataset and verify the export excludes it
Explanation
Correct Answer
D:
Create an unrelated dataset and verify the export excludes it is a direct acceptance test of the path implicated by audit log shows query scope expanded after a new dataset was added. It checks that the original risk (a nightly bq CLI export uses wildcard datasets and leaks unrelated faculty research) has actually been controlled and tests that permitted operations still meet the requirement that exports must include only approved course statistics.
Incorrect Answers
A:
The alternative “Deploy export script with bounded project context and approved schedules” documents change execution but not its operational acceptance result. In the reported observation, audit log shows query scope expanded after a new dataset was added. The option “Create an unrelated dataset and verify the export excludes it” instead addresses the requested acceptance verification.
B:
The alternative “Identify exact dataset boundaries, IAM roles and query requirements” collects prechange evidence without confirming that the revised service meets its outcome. In the reported observation, audit log shows query scope expanded after a new dataset was added. The option “Create an unrelated dataset and verify the export excludes it” instead addresses the requested acceptance verification.
C:
Broad admin rights violate the limited read-only extraction purpose. For this question, it is not a defensible choice for acceptance verification and it does not satisfy the requirement that exports must include only approved course statistics.
Question 908
At University financial aid, the support access audit workload has a documented risk: a troubleshooting engineer needs to inspect errors but receives project Owner privileges. The review found that permission request includes unnecessary user data access. The architecture must also ensure that incident support must follow least privilege and logged access. What enduring Google Cloud design should the team choose now that the failure mode is understood?
- Give all temporary support contractors permanent Owner access
- Determine exact debug resources, logs and role requirements
- Watch stale support grants and unauthorized data reads
- Create scoped temporary diagnostic roles and privileged session recording
Explanation
Correct Answer
D:
Create scoped temporary diagnostic roles and privileged session recording targets the underlying issue (a troubleshooting engineer needs to inspect errors but receives project Owner privileges) rather than merely reporting symptoms. It also preserves the required outcome (incident support must follow least privilege and logged access), which a target-state design must satisfy.
Incorrect Answers
A:
Permanent broad privileges are not justified by a narrow debugging need. For this question, it is not a defensible choice for architecture selection and it does not satisfy the requirement that incident support must follow least privilege and logged access.
B:
The alternative “Determine exact debug resources, logs and role requirements” investigates the cause without specifying the durable architecture to resolve it. In the reported observation, permission request includes unnecessary user data access. The option “Create scoped temporary diagnostic roles and privileged session recording” instead addresses the requested architecture selection.
C:
The alternative “Watch stale support grants and unauthorized data reads” measures results without defining the preventive design now required. In the reported observation, permission request includes unnecessary user data access. The option “Create scoped temporary diagnostic roles and privileged session recording” instead addresses the requested architecture selection.
Question 909
The accessibility acceptance environment at City library data portal faces this issue: a redesigned public portal uses keyboard-inaccessible chart filters. Supporting evidence shows that accessibility audit finds focus cannot reach essential search fields. The architect must preserve the requirement that government content must be accessible to assistive technology users. Go-live and initial tests passed. Which ongoing measure is most useful to catch control drift or regression?
- Complete search using keyboard and screen reader without mouse interaction
- Track accessibility regressions and user support reports
- Include accessible components and automated/manual quality criteria in release requirements
- Approve launch solely because automated unit tests passed
Explanation
Correct Answer
B:
Track accessibility regressions and user support reports reveals recurrence, drift or changing dependencies after launch. It provides a repeatable signal for the risk (a redesigned public portal uses keyboard-inaccessible chart filters) rather than treating the first successful test as permanent assurance.
Incorrect Answers
A:
The alternative “Complete search using keyboard and screen reader without mouse interaction” is a one-time acceptance result rather than ongoing operational assurance. In the reported observation, accessibility audit finds focus cannot reach essential search fields. The option “Track accessibility regressions and user support reports” instead addresses the requested ongoing operational assurance.
C:
The alternative “Include accessible components and automated/manual quality criteria in release requirements” selects a design, but does not provide the ongoing signal required to detect drift. In the reported observation, accessibility audit finds focus cannot reach essential search fields. The option “Track accessibility regressions and user support reports” instead addresses the requested ongoing operational assurance.
D:
Unit tests do not capture all interactive accessibility barriers. For this question, it is not a defensible choice for ongoing operational assurance and it does not satisfy the requirement that government content must be accessible to assistive technology users.
Question 910
Consider Municipal permit office and its digital permit filing workload. The issue is that stakeholders propose a simple document store without cross-department approval traceability; the current observation is that a discovery workshop finds six approval signatures and statutory retention deadlines. The team must ensure that citizens must see accurate permit status across agencies. Which acceptance test should demonstrate both protection against the failure and continued authorized service?
- Phase department onboarding with dual reporting and reconciliation
- Replace all human approvals with a single anonymous automated approval
- Verify every signed approval is linked to one permit version and visible to authorized users
- Document approval workflow, record lifecycle, access roles and integration requirements
Explanation
Correct Answer
C:
Verify every signed approval is linked to one permit version and visible to authorized users is a direct acceptance test of the path implicated by a discovery workshop finds six approval signatures and statutory retention deadlines. It checks that the original risk (stakeholders propose a simple document store without cross-department approval traceability) has actually been controlled and tests that permitted operations still meet the requirement that citizens must see accurate permit status across agencies.
Incorrect Answers
A:
The alternative “Phase department onboarding with dual reporting and reconciliation” documents change execution but not its operational acceptance result. In the reported observation, a discovery workshop finds six approval signatures and statutory retention deadlines. The option “Verify every signed approval is linked to one permit version and visible to authorized users” instead addresses the requested acceptance verification.
B:
An unaccountable shortcut cannot satisfy multi-agency statutory approval. For this question, it is not a defensible choice for acceptance verification and it does not satisfy the requirement that citizens must see accurate permit status across agencies.
D:
The alternative “Document approval workflow, record lifecycle, access roles and integration requirements” collects prechange evidence without confirming that the revised service meets its outcome. In the reported observation, a discovery workshop finds six approval signatures and statutory retention deadlines. The option “Verify every signed approval is linked to one permit version and visible to authorized users” instead addresses the requested acceptance verification.