Google Professional Cloud Architect Practice Test 39
Question 380
At City library data portal, the Google Cloud design for accessibility acceptance is affected because a redesigned public portal uses keyboard-inaccessible chart filters. A review found that accessibility audit finds focus cannot reach essential search fields. The proposed response must account for the requirement that government content must be accessible to assistive technology users. Which controlled implementation should be used to preserve operational continuity during the change?
- Track accessibility regressions and user support reports
- Release fixes behind user-testing feedback and rollback
- Approve launch solely because automated unit tests passed
- Include accessible components and automated/manual quality criteria in release requirements
Explanation
Correct Answer
B:
Release fixes behind user-testing feedback and rollback is a controlled means to introduce the approved architecture while meeting the requirement that government content must be accessible to assistive technology users. Staged delivery, attributable change evidence and rollback guard against an otherwise avoidable service disruption.
Incorrect Answers
A:
The alternative “Track accessibility regressions and user support reports” addresses later monitoring rather than the actual controlled deployment. In the reported observation, accessibility audit finds focus cannot reach essential search fields. The option “Release fixes behind user-testing feedback and rollback” instead addresses the requested controlled implementation.
C:
Unit tests do not capture all interactive accessibility barriers. For this question, it is not a defensible choice for controlled implementation and it does not satisfy the requirement that government content must be accessible to assistive technology users.
D:
The alternative “Include accessible components and automated/manual quality criteria in release requirements” describes the chosen design, but not how it is introduced safely. In the reported observation, accessibility audit finds focus cannot reach essential search fields. The option “Release fixes behind user-testing feedback and rollback” instead addresses the requested controlled implementation.
Question 381
At Education analytics cooperative, the data pipeline incident triage workload has a documented risk: daily analytics job silently drops events with previously unseen nullable fields. The review found that Dataflow error logs show schema deserialization failures after partner update. The architecture must also ensure that approved learning dashboards require source-to-target reconciliation. Select TWO complementary actions: one durable target architecture and one controlled way to introduce it while preserving service. Which TWO are best?
- Discard all unrecognized events without alerting
- Compare payload schema versions, dead-letter events and parsing code
- Define versioned schemas with controlled evolution and quarantined invalid events
- Monitor schema rejection and lag by producer
- Deploy tolerant parsers in staged pipelines with recovery of quarantined records
Explanation
Correct Answers
C:
Define versioned schemas with controlled evolution and quarantined invalid events targets the underlying issue (daily analytics job silently drops events with previously unseen nullable fields) rather than merely reporting symptoms. It also preserves the required outcome (approved learning dashboards require source-to-target reconciliation), which a target-state design must satisfy. This supplies the lasting-control half of the two-action package.
E:
Deploy tolerant parsers in staged pipelines with recovery of quarantined records is a controlled means to introduce the approved architecture while meeting the requirement that approved learning dashboards require source-to-target reconciliation. 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
A:
Silent discards create false analytics that appear complete. The shortcut does not supply the requested design and safe rollout for data pipeline incident triage, where Dataflow error logs show schema deserialization failures after partner update.
B:
This diagnoses the original issue (daily analytics job silently drops events with previously unseen nullable fields), but the TWO requested actions are the target architecture and its implementation; discovery was already completed.
D:
Monitoring data pipeline incident triage for future drift is useful after the rollout, but it does not specify the requested target control and safe implementation for this case: daily analytics job silently drops events with previously unseen nullable fields.
Question 382
Consider Global fleet analytics company and its connected vehicle insight platform workload. The issue is that management expects ten times more telemetry after a merger but the platform allocates fixed partitions; the current observation is that forecast shows devices and event attributes will expand across regions. The team must ensure that existing fleet alerts cannot lose events during schema evolution. The target architecture is approved; execution without disruption is the remaining decision. Which rollout action is best?
- Roll out schema evolution with dual parsers and backward-compatible records
- Design partitioned event contracts and independently scalable analytical consumers
- Watch partition hotspots, obsolete schemas and capacity forecasts
- Change all event schemas simultaneously without versioning
Explanation
Correct Answer
A:
Roll out schema evolution with dual parsers and backward-compatible records is a controlled means to introduce the approved architecture while meeting the requirement that existing fleet alerts cannot lose events during schema evolution. Staged delivery, attributable change evidence and rollback guard against an otherwise avoidable service disruption.
Incorrect Answers
B:
The alternative “Design partitioned event contracts and independently scalable analytical consumers” describes the chosen design, but not how it is introduced safely. In the reported observation, forecast shows devices and event attributes will expand across regions. The option “Roll out schema evolution with dual parsers and backward-compatible records” instead addresses the requested controlled implementation.
C:
The alternative “Watch partition hotspots, obsolete schemas and capacity forecasts” addresses later monitoring rather than the actual controlled deployment. In the reported observation, forecast shows devices and event attributes will expand across regions. The option “Roll out schema evolution with dual parsers and backward-compatible records” instead addresses the requested controlled implementation.
D:
A schema cutover without compatibility can strand existing devices. For this question, it is not a defensible choice for controlled implementation and it does not satisfy the requirement that existing fleet alerts cannot lose events during schema evolution.
Question 383
At Public university registrar, the student-data migration workload has a documented risk: bulk import from an aging database overwrites later updates during the cutover. The review found that comparison shows primary-key collisions between initial exports and ongoing edits. The architecture must also ensure that graduation records must be correct at the moment of cutover. The architecture review is paused until the team can prove the real failure path. Which investigation should happen first?
- Execute synchronized cutover after low-lag catch-up and final write-freeze
- Import the original snapshot only and ignore writes after its extraction
- Map CDC support, key semantics, update order and rollback window
- Plan full-load plus change-data capture with deterministic conflict handling
Explanation
Correct Answer
C:
Map CDC support, key semantics, update order and rollback window is the appropriate diagnostic because the observation—comparison shows primary-key collisions between initial exports and ongoing edits—does not by itself establish every dependency, authorization rule or failure boundary. The findings explain the reported risk (bulk import from an aging database overwrites later updates during the cutover) before the design is approved.
Incorrect Answers
A:
The alternative “Execute synchronized cutover after low-lag catch-up and final write-freeze” is rollout work that may modify the wrong dependency before root-cause evidence is collected. In the reported observation, comparison shows primary-key collisions between initial exports and ongoing edits. The option “Map CDC support, key semantics, update order and rollback window” instead addresses the requested root-cause evidence.
B:
A stale snapshot loses legitimate updates made during the move. For this question, it is not a defensible choice for root-cause evidence and it does not satisfy the requirement that graduation records must be correct at the moment of cutover.
D:
The alternative “Plan full-load plus change-data capture with deterministic conflict handling” is target architecture selection; it does not establish why the observed path failed. In the reported observation, comparison shows primary-key collisions between initial exports and ongoing edits. The option “Map CDC support, key semantics, update order and rollback window” instead addresses the requested root-cause evidence.
Question 384
At Healthcare AI supplier, the Google Cloud design for GKE node-upgrade safety is affected because a node upgrade evicts all inference replicas because they share one failure domain. A review found that cluster events show two replicas pinned to the same maintenance pool. The proposed response must account for the requirement that inference must remain available during security patching. A deployment ticket has been closed. What direct evidence would justify solution signoff?
- Inspect pod affinity, disruption budgets and node-pool placement
- Set replica count to one and disable disruption checks
- Execute a controlled node drain and verify inference availability
- Move one replica and drain one node per upgrade wave
Explanation
Correct Answer
C:
Execute a controlled node drain and verify inference availability is a direct acceptance test of the path implicated by cluster events show two replicas pinned to the same maintenance pool. It checks that the original risk (a node upgrade evicts all inference replicas because they share one failure domain) has actually been controlled and tests that permitted operations still meet the requirement that inference must remain available during security patching.
Incorrect Answers
A:
The alternative “Inspect pod affinity, disruption budgets and node-pool placement” collects prechange evidence without confirming that the revised service meets its outcome. In the reported observation, cluster events show two replicas pinned to the same maintenance pool. The option “Execute a controlled node drain and verify inference availability” instead addresses the requested acceptance verification.
B:
A single replica cannot tolerate a node drain. For this question, it is not a defensible choice for acceptance verification and it does not satisfy the requirement that inference must remain available during security patching.
D:
The alternative “Move one replica and drain one node per upgrade wave” documents change execution but not its operational acceptance result. In the reported observation, cluster events show two replicas pinned to the same maintenance pool. The option “Execute a controlled node drain and verify inference availability” instead addresses the requested acceptance verification.
Question 385
The predictive maintenance model environment at Solar equipment company faces this issue: trained model artifacts are deployed manually and inference behavior drifts across regions. Supporting evidence shows that model logs show differing preprocessing versions between plant gateways. The architect must preserve the requirement that anomaly decisions must have lineage and allow rollback. What test must pass before the business accepts the remediated architecture?
- Edit model weights directly inside each running inference container
- Deploy canary model versions with signed artifacts and rollback gate
- Compare model outputs on held-out plant data across serving regions
- Inventory feature pipeline versions, model registry lineage and inference paths
Explanation
Correct Answer
C:
Compare model outputs on held-out plant data across serving regions is a direct acceptance test of the path implicated by model logs show differing preprocessing versions between plant gateways. It checks that the original risk (trained model artifacts are deployed manually and inference behavior drifts across regions) has actually been controlled and tests that permitted operations still meet the requirement that anomaly decisions must have lineage and allow rollback.
Incorrect Answers
A:
Unversioned manual edits destroy reproducibility and rollback. For this question, it is not a defensible choice for acceptance verification and it does not satisfy the requirement that anomaly decisions must have lineage and allow rollback.
B:
The alternative “Deploy canary model versions with signed artifacts and rollback gate” documents change execution but not its operational acceptance result. In the reported observation, model logs show differing preprocessing versions between plant gateways. The option “Compare model outputs on held-out plant data across serving regions” instead addresses the requested acceptance verification.
D:
The alternative “Inventory feature pipeline versions, model registry lineage and inference paths” collects prechange evidence without confirming that the revised service meets its outcome. In the reported observation, model logs show differing preprocessing versions between plant gateways. The option “Compare model outputs on held-out plant data across serving regions” instead addresses the requested acceptance verification.
Question 386
Consider Retail healthcare supplier and its Cloud Run canary release workload. The issue is that a monolithic new API revision receives 100% of traffic and breaks patient orders; the current observation is that release journal shows no canary or rollback criterion. The team must ensure that approved orders must continue during application upgrades. The team has completed discovery. Which preventive or corrective architecture is most appropriate?
- Monitor revision-specific errors and rollback decision latency
- Replace the production revision all at once and delete the prior image
- Design progressive Cloud Run revisions with measurable abort conditions
- Analyze failure budget, revision behavior and traffic splitting needs
Explanation
Correct Answer
C:
Design progressive Cloud Run revisions with measurable abort conditions targets the underlying issue (a monolithic new API revision receives 100% of traffic and breaks patient orders) rather than merely reporting symptoms. It also preserves the required outcome (approved orders must continue during application upgrades), which a target-state design must satisfy.
Incorrect Answers
A:
The alternative “Monitor revision-specific errors and rollback decision latency” measures results without defining the preventive design now required. In the reported observation, release journal shows no canary or rollback criterion. The option “Design progressive Cloud Run revisions with measurable abort conditions” instead addresses the requested architecture selection.
B:
An irreversible full cutover magnifies deployment failure. For this question, it is not a defensible choice for architecture selection and it does not satisfy the requirement that approved orders must continue during application upgrades.
D:
The alternative “Analyze failure budget, revision behavior and traffic splitting needs” investigates the cause without specifying the durable architecture to resolve it. In the reported observation, release journal shows no canary or rollback criterion. The option “Design progressive Cloud Run revisions with measurable abort conditions” instead addresses the requested architecture selection.
Question 387
The SLI boundary selection environment at Regional bank mortgage service faces this issue: frontend availability dashboard shows 100% while database commit errors abort applications. Supporting evidence shows that synthetic API probes use a cached read endpoint only. The architect must preserve the requirement that home-loan approvals require successful authorized durable writes. Go-live and initial tests passed. Which ongoing measure is most useful to catch control drift or regression?
- Set an end-to-end write-success SLI with trace and durable-storage evidence
- Force a database write failure and confirm SLI/alert reflects unsuccessful approvals
- Monitor failed durable writes and request latency by product
- Use only HTTP ingress uptime as the loan approval SLI
Explanation
Correct Answer
C:
Monitor failed durable writes and request latency by product reveals recurrence, drift or changing dependencies after launch. It provides a repeatable signal for the risk (frontend availability dashboard shows 100% while database commit errors abort applications) rather than treating the first successful test as permanent assurance.
Incorrect Answers
A:
The alternative “Set an end-to-end write-success SLI with trace and durable-storage evidence” selects a design, but does not provide the ongoing signal required to detect drift. In the reported observation, synthetic API probes use a cached read endpoint only. The option “Monitor failed durable writes and request latency by product” instead addresses the requested ongoing operational assurance.
B:
The alternative “Force a database write failure and confirm SLI/alert reflects unsuccessful approvals” is a one-time acceptance result rather than ongoing operational assurance. In the reported observation, synthetic API probes use a cached read endpoint only. The option “Monitor failed durable writes and request latency by product” instead addresses the requested ongoing operational assurance.
D:
A reachable proxy does not prove committed financial transactions. For this question, it is not a defensible choice for ongoing operational assurance and it does not satisfy the requirement that home-loan approvals require successful authorized durable writes.
Question 388
National weather agency plans to update alert dissemination. Investigators found that metrics reveal static instances hit connection limits while message publication remains healthy, consistent with the reported problem that the weather alert app must handle a sudden 30-fold audience jump when storms hit. The response must satisfy the requirement that government alerts must be delivered promptly during severe weather. The team needs an operations signal rather than another design or one-time test. What should it track?
- Disable rate and capacity limits without testing downstream quotas
- Generate synthetic 30x fanout and confirm alert freshness and error budgets
- Separate publish from fan-out with managed scaling and geographic caching
- Track autoscaler quotas, queue backlog and notification latency
Explanation
Correct Answer
D:
Track autoscaler quotas, queue backlog and notification latency reveals recurrence, drift or changing dependencies after launch. It provides a repeatable signal for the risk (the weather alert app must handle a sudden 30-fold audience jump when storms hit) rather than treating the first successful test as permanent assurance.
Incorrect Answers
A:
Unbounded fanout can exhaust dependencies rather than improving availability. For this question, it is not a defensible choice for ongoing operational assurance and it does not satisfy the requirement that government alerts must be delivered promptly during severe weather.
B:
The alternative “Generate synthetic 30x fanout and confirm alert freshness and error budgets” is a one-time acceptance result rather than ongoing operational assurance. In the reported observation, metrics reveal static instances hit connection limits while message publication remains healthy. The option “Track autoscaler quotas, queue backlog and notification latency” instead addresses the requested ongoing operational assurance.
C:
The alternative “Separate publish from fan-out with managed scaling and geographic caching” selects a design, but does not provide the ongoing signal required to detect drift. In the reported observation, metrics reveal static instances hit connection limits while message publication remains healthy. The option “Track autoscaler quotas, queue backlog and notification latency” instead addresses the requested ongoing operational assurance.
Question 389
Community bank plans to update SQL query optimization. Investigators found that EXPLAIN plans show full scans on a growing application table, consistent with the reported problem that a loan search endpoint slowed after adding optional fuzzy-name filters. The response must satisfy the requirement that loan updates must remain strongly consistent. The security and reliability teams dispute the root cause. Which evidence-gathering action is justified now?
- Design appropriate indexes and search pattern without changing transactional semantics
- Inspect query plans, index selectivity and p95 response decomposition
- Add more web servers without addressing the repeated full SQL scans
- Deploy candidate indexes in off-peak maintenance with rollback
Explanation
Correct Answer
B:
Inspect query plans, index selectivity and p95 response decomposition is the appropriate diagnostic because the observation—EXPLAIN plans show full scans on a growing application table—does not by itself establish every dependency, authorization rule or failure boundary. The findings explain the reported risk (a loan search endpoint slowed after adding optional fuzzy-name filters) before the design is approved.
Incorrect Answers
A:
The alternative “Design appropriate indexes and search pattern without changing transactional semantics” is target architecture selection; it does not establish why the observed path failed. In the reported observation, EXPLAIN plans show full scans on a growing application table. The option “Inspect query plans, index selectivity and p95 response decomposition” instead addresses the requested root-cause evidence.
C:
A database plan problem will persist across extra web instances. For this question, it is not a defensible choice for root-cause evidence and it does not satisfy the requirement that loan updates must remain strongly consistent.
D:
The alternative “Deploy candidate indexes in off-peak maintenance with rollback” is rollout work that may modify the wrong dependency before root-cause evidence is collected. In the reported observation, EXPLAIN plans show full scans on a growing application table. The option “Inspect query plans, index selectivity and p95 response decomposition” instead addresses the requested root-cause evidence.
Question 390
At Automotive safety modeling startup, the reinforcement model evaluation workload has a documented risk: engineering wants to deploy a new model before proving corner-case behavior. The review found that evaluation report covers average accuracy but no rare safety-critical maneuvers. The architecture must also ensure that safety review requires rigorous offline evaluation and rollback. The architecture review is paused until the team can prove the real failure path. Which investigation should happen first?
- Introduce shadow inference before any controlled production promotion
- Approve the model solely because aggregate accuracy rose one percent
- Design model lifecycle gates with offline evaluation and isolated acceptance
- Define rare-event evaluation cases, adversarial tests and traceability
Explanation
Correct Answer
D:
Define rare-event evaluation cases, adversarial tests and traceability is the appropriate diagnostic because the observation—evaluation report covers average accuracy but no rare safety-critical maneuvers—does not by itself establish every dependency, authorization rule or failure boundary. The findings explain the reported risk (engineering wants to deploy a new model before proving corner-case behavior) before the design is approved.
Incorrect Answers
A:
The alternative “Introduce shadow inference before any controlled production promotion” is rollout work that may modify the wrong dependency before root-cause evidence is collected. In the reported observation, evaluation report covers average accuracy but no rare safety-critical maneuvers. The option “Define rare-event evaluation cases, adversarial tests and traceability” instead addresses the requested root-cause evidence.
B:
Mean accuracy can hide unacceptable safety-regression tails. For this question, it is not a defensible choice for root-cause evidence and it does not satisfy the requirement that safety review requires rigorous offline evaluation and rollback.
C:
The alternative “Design model lifecycle gates with offline evaluation and isolated acceptance” is target architecture selection; it does not establish why the observed path failed. In the reported observation, evaluation report covers average accuracy but no rare safety-critical maneuvers. The option “Define rare-event evaluation cases, adversarial tests and traceability” instead addresses the requested root-cause evidence.