Google Professional Cloud Architect Practice Test 98
Question 970
Case study: Medical device logistics vendor: Windows application portfolio
Medical device logistics vendor relies on Windows application portfolio to deliver its operational commitments. The current architectural problem is that a vendor application requires proprietary Windows integration and cannot be modified until 2028. The team has recorded this evidence: application inventory confirms hard-coded Windows service dependency and vendor support conditions. Any proposed response must preserve the requirement that device dispatch must maintain support entitlement and tested restoration.
The Windows application portfolio environment at Medical device logistics vendor faces this issue: a vendor application requires proprietary Windows integration and cannot be modified until 2028. Supporting evidence shows that application inventory confirms hard-coded Windows service dependency and vendor support conditions. The architect must preserve the requirement that device dispatch must maintain support entitlement and tested restoration. What test must pass before the business accepts the remediated architecture?
- Convert the unsupported proprietary service directly to Cloud Run with no rewrite
- Create reproducible images and migrate one warehouse environment with rollback
- Run vendor certification tests and warehouse dispatch transactions on restored instances
- Assess OS, licensing, database and vendor certification constraints
Explanation
Correct Answer
C:
Run vendor certification tests and warehouse dispatch transactions on restored instances is a direct acceptance test of the path implicated by application inventory confirms hard-coded Windows service dependency and vendor support conditions. It checks that the original risk (a vendor application requires proprietary Windows integration and cannot be modified until 2028) has actually been controlled and tests that permitted operations still meet the requirement that device dispatch must maintain support entitlement and tested restoration.
Incorrect Answers
A:
A Windows-service binary does not become a stateless Linux container automatically. For this question, it is not a defensible choice for acceptance verification and it does not satisfy the requirement that device dispatch must maintain support entitlement and tested restoration.
B:
The alternative “Create reproducible images and migrate one warehouse environment with rollback” documents change execution but not its operational acceptance result. In the reported observation, application inventory confirms hard-coded Windows service dependency and vendor support conditions. The option “Run vendor certification tests and warehouse dispatch transactions on restored instances” instead addresses the requested acceptance verification.
D:
The alternative “Assess OS, licensing, database and vendor certification constraints” collects prechange evidence without confirming that the revised service meets its outcome. In the reported observation, application inventory confirms hard-coded Windows service dependency and vendor support conditions. The option “Run vendor certification tests and warehouse dispatch transactions on restored instances” instead addresses the requested acceptance verification.
Question 971
At Public health research hub, the private SaaS producer endpoint workload has a documented risk: consumer projects connect over public load balancers despite producer and consumers residing in Google Cloud. The review found that traceroute confirms data traverses a public endpoint with IP-based allowlists. The architecture must also ensure that approved projects require internal-only access and separate VPC ownership. The architecture review is paused until the team can prove the real failure path. Which investigation should happen first?
- Grant approved consumer endpoint access and update internal DNS in phases
- Enable public access with a broad CIDR allowlist and call it private
- Map service producer/consumer boundaries and traffic direction
- Expose the private producer service through Private Service Connect endpoints
Explanation
Correct Answer
C:
Map service producer/consumer boundaries and traffic direction is the appropriate diagnostic because the observation—traceroute confirms data traverses a public endpoint with IP-based allowlists—does not by itself establish every dependency, authorization rule or failure boundary. The findings explain the reported risk (consumer projects connect over public load balancers despite producer and consumers residing in Google Cloud) before the design is approved.
Incorrect Answers
A:
The alternative “Grant approved consumer endpoint access and update internal DNS in phases” is rollout work that may modify the wrong dependency before root-cause evidence is collected. In the reported observation, traceroute confirms data traverses a public endpoint with IP-based allowlists. The option “Map service producer/consumer boundaries and traffic direction” instead addresses the requested root-cause evidence.
B:
Public routing and wide CIDR access do not meet a private-service boundary. For this question, it is not a defensible choice for root-cause evidence and it does not satisfy the requirement that approved projects require internal-only access and separate VPC ownership.
D:
The alternative “Expose the private producer service through Private Service Connect endpoints” is target architecture selection; it does not establish why the observed path failed. In the reported observation, traceroute confirms data traverses a public endpoint with IP-based allowlists. The option “Map service producer/consumer boundaries and traffic direction” instead addresses the requested root-cause evidence.
Question 972
Local telecom provider plans to update VMware-based customer management. Investigators found that migration inventory shows tight network coupling and unsupported guest customization, consistent with the reported problem that team wants to migrate hundreds of VMware virtual machines with almost no application changes. The response must satisfy the requirement that a contract requires a one-year low-risk transitional hosting period. The approved change has been applied. Which validation provides the clearest evidence the original problem is fixed?
- Verify application behavior, license constraints and VM restore on target
- Delete the existing hypervisor configurations before verifying guest compatibility
- Pilot a VM cohort with private connectivity and validated operational ownership
- Map VMware dependencies, supported software versions and network IP collisions
Explanation
Correct Answer
A:
Verify application behavior, license constraints and VM restore on target is a direct acceptance test of the path implicated by migration inventory shows tight network coupling and unsupported guest customization. It checks that the original risk (team wants to migrate hundreds of VMware virtual machines with almost no application changes) has actually been controlled and tests that permitted operations still meet the requirement that a contract requires a one-year low-risk transitional hosting period.
Incorrect Answers
B:
Unplanned transformation creates unacceptable migration risk. For this question, it is not a defensible choice for acceptance verification and it does not satisfy the requirement that a contract requires a one-year low-risk transitional hosting period.
C:
The alternative “Pilot a VM cohort with private connectivity and validated operational ownership” documents change execution but not its operational acceptance result. In the reported observation, migration inventory shows tight network coupling and unsupported guest customization. The option “Verify application behavior, license constraints and VM restore on target” instead addresses the requested acceptance verification.
D:
The alternative “Map VMware dependencies, supported software versions and network IP collisions” collects prechange evidence without confirming that the revised service meets its outcome. In the reported observation, migration inventory shows tight network coupling and unsupported guest customization. The option “Verify application behavior, license constraints and VM restore on target” instead addresses the requested acceptance verification.
Question 973
At Emergency medical scheduling startup, the feature-flag rollback workload has a documented risk: a risky appointment algorithm is released to all users before its impacts can be measured. The review found that pilot notes show accepted bookings are reallocated unexpectedly. The architecture must also ensure that existing appointments must remain stable during launch. What follow-up metric or alert will identify emerging failures after the initial acceptance test?
- Watch cohort-specific booking errors and rollback latency
- Verify rollback keeps existing bookings unchanged and future requests safe
- Use controlled feature flags with experiment guards and durable booking invariants
- Modify all stored appointments irreversibly when enabling the feature
Explanation
Correct Answer
A:
Watch cohort-specific booking errors and rollback latency reveals recurrence, drift or changing dependencies after launch. It provides a repeatable signal for the risk (a risky appointment algorithm is released to all users before its impacts can be measured) rather than treating the first successful test as permanent assurance.
Incorrect Answers
B:
The alternative “Verify rollback keeps existing bookings unchanged and future requests safe” is a one-time acceptance result rather than ongoing operational assurance. In the reported observation, pilot notes show accepted bookings are reallocated unexpectedly. The option “Watch cohort-specific booking errors and rollback latency” instead addresses the requested ongoing operational assurance.
C:
The alternative “Use controlled feature flags with experiment guards and durable booking invariants” selects a design, but does not provide the ongoing signal required to detect drift. In the reported observation, pilot notes show accepted bookings are reallocated unexpectedly. The option “Watch cohort-specific booking errors and rollback latency” instead addresses the requested ongoing operational assurance.
D:
Feature flags cannot undo unprotected database side effects. For this question, it is not a defensible choice for ongoing operational assurance and it does not satisfy the requirement that existing appointments must remain stable during launch.
Question 974
Regional financial adviser plans to update private research assistant. Investigators found that demo agent follows instructions embedded in retrieved research PDFs, consistent with the reported problem that employees want document-grounded answers without granting a chatbot access to privileged trading APIs. The response must satisfy the requirement that approved internal reports must be cited and trading authorization separate. The approved change has been applied. Which validation provides the clearest evidence the original problem is fixed?
- Inject malicious retrieved instructions and verify no trading call executes
- Map trusted retrieved context and the tool-authorization boundary
- Treat every retrieved document as an instruction from the security administrator
- Introduce approved corpus sources with prompt-injection tests and read-only permissions
Explanation
Correct Answer
A:
Inject malicious retrieved instructions and verify no trading call executes is a direct acceptance test of the path implicated by demo agent follows instructions embedded in retrieved research PDFs. It checks that the original risk (employees want document-grounded answers without granting a chatbot access to privileged trading APIs) has actually been controlled and tests that permitted operations still meet the requirement that approved internal reports must be cited and trading authorization separate.
Incorrect Answers
B:
The alternative “Map trusted retrieved context and the tool-authorization boundary” collects prechange evidence without confirming that the revised service meets its outcome. In the reported observation, demo agent follows instructions embedded in retrieved research PDFs. The option “Inject malicious retrieved instructions and verify no trading call executes” instead addresses the requested acceptance verification.
C:
Untrusted document content cannot grant privileged tool authority. For this question, it is not a defensible choice for acceptance verification and it does not satisfy the requirement that approved internal reports must be cited and trading authorization separate.
D:
The alternative “Introduce approved corpus sources with prompt-injection tests and read-only permissions” documents change execution but not its operational acceptance result. In the reported observation, demo agent follows instructions embedded in retrieved research PDFs. The option “Inject malicious retrieved instructions and verify no trading call executes” instead addresses the requested acceptance verification.
Question 975
International hotel chain plans to update central incident notification API. Investigators found that test customers see duplicate room-change notifications, consistent with the reported problem that operations migrates a notification service before updating caller authentication and message retry rules. The response must satisfy the requirement that customer booking updates must remain unique and attributable. The security and reliability teams dispute the root cause. Which evidence-gathering action is justified now?
- Move partners through versioned API deployment with replayable messages
- Disable all retries and discard messages when the first send fails
- Review retry, authentication and idempotency semantics across partners
- Design idempotent notification delivery with scoped API identities
Explanation
Correct Answer
C:
Review retry, authentication and idempotency semantics across partners is the appropriate diagnostic because the observation—test customers see duplicate room-change notifications—does not by itself establish every dependency, authorization rule or failure boundary. The findings explain the reported risk (operations migrates a notification service before updating caller authentication and message retry rules) before the design is approved.
Incorrect Answers
A:
The alternative “Move partners through versioned API deployment with replayable messages” is rollout work that may modify the wrong dependency before root-cause evidence is collected. In the reported observation, test customers see duplicate room-change notifications. The option “Review retry, authentication and idempotency semantics across partners” instead addresses the requested root-cause evidence.
B:
At-most-once loss does not meet delivery commitments. For this question, it is not a defensible choice for root-cause evidence and it does not satisfy the requirement that customer booking updates must remain unique and attributable.
D:
The alternative “Design idempotent notification delivery with scoped API identities” is target architecture selection; it does not establish why the observed path failed. In the reported observation, test customers see duplicate room-change notifications. The option “Review retry, authentication and idempotency semantics across partners” instead addresses the requested root-cause evidence.
Question 976
At City library data portal, the accessibility acceptance workload has a documented risk: a redesigned public portal uses keyboard-inaccessible chart filters. The review found that accessibility audit finds focus cannot reach essential search fields. The architecture must also ensure that government content must be accessible to assistive technology users. Which post-change test most reliably distinguishes a functioning control from a merely successful installation?
- Release fixes behind user-testing feedback and rollback
- Approve launch solely because automated unit tests passed
- Complete search using keyboard and screen reader without mouse interaction
- Review key workflows, keyboard navigation and applicable accessibility criteria
Explanation
Correct Answer
C:
Complete search using keyboard and screen reader without mouse interaction is a direct acceptance test of the path implicated by accessibility audit finds focus cannot reach essential search fields. It checks that the original risk (a redesigned public portal uses keyboard-inaccessible chart filters) has actually been controlled and tests that permitted operations still meet the requirement that government content must be accessible to assistive technology users.
Incorrect Answers
A:
The alternative “Release fixes behind user-testing feedback and rollback” documents change execution but not its operational acceptance result. In the reported observation, accessibility audit finds focus cannot reach essential search fields. The option “Complete search using keyboard and screen reader without mouse interaction” instead addresses the requested acceptance verification.
B:
Unit tests do not capture all interactive accessibility barriers. For this question, it is not a defensible choice for acceptance verification and it does not satisfy the requirement that government content must be accessible to assistive technology users.
D:
The alternative “Review key workflows, keyboard navigation and applicable accessibility criteria” collects prechange evidence without confirming that the revised service meets its outcome. In the reported observation, accessibility audit finds focus cannot reach essential search fields. The option “Complete search using keyboard and screen reader without mouse interaction” instead addresses the requested acceptance verification.
Question 977
The PCI data encryption separation environment at Financial consumer lender faces this issue: application developers can read a cardholder encryption key and production data. Supporting evidence shows that role review shows one service account can decrypt and administer Cloud KMS. The architect must preserve the requirement that PCI program requires separation of key administration and data processing. Go-live and initial tests passed. Which ongoing measure is most useful to catch control drift or regression?
- Monitor key-use logs, privileged changes and quarterly access reviews
- Apply separate IAM roles for key administration and cryptographic use, with CMEK where supported
- Set a shared spreadsheet password on exported plaintext cards
- Verify developers cannot decrypt card records or destroy keys
Explanation
Correct Answer
A:
Monitor key-use logs, privileged changes and quarterly access reviews reveals recurrence, drift or changing dependencies after launch. It provides a repeatable signal for the risk (application developers can read a cardholder encryption key and production data) rather than treating the first successful test as permanent assurance.
Incorrect Answers
B:
The alternative “Apply separate IAM roles for key administration and cryptographic use, with CMEK where supported” selects a design, but does not provide the ongoing signal required to detect drift. In the reported observation, role review shows one service account can decrypt and administer Cloud KMS. The option “Monitor key-use logs, privileged changes and quarterly access reviews” instead addresses the requested ongoing operational assurance.
C:
A shared password does not meet secure key custody requirements. For this question, it is not a defensible choice for ongoing operational assurance and it does not satisfy the requirement that PCI program requires separation of key administration and data processing.
D:
The alternative “Verify developers cannot decrypt card records or destroy keys” is a one-time acceptance result rather than ongoing operational assurance. In the reported observation, role review shows one service account can decrypt and administer Cloud KMS. The option “Monitor key-use logs, privileged changes and quarterly access reviews” instead addresses the requested ongoing operational assurance.
Question 978
The automated infrastructure catalog environment at Regional newspaper platform faces this issue: each team manually creates different VM images for the same application tier. Supporting evidence shows that asset review shows inconsistent IAM, patch settings and logging between regions. The architect must preserve the requirement that new sites must have consistent repeatable baseline controls. Which migration or rollout step best manages risk, dependencies and reversibility?
- Let each engineer edit production resources manually without recording changes
- Publish reusable Terraform modules with policy tests and documented ownership
- Migrate one region to versioned modules and controlled state storage
- Monitor module drift, change approvals and obsolete versions
Explanation
Correct Answer
C:
Migrate one region to versioned modules and controlled state storage is a controlled means to introduce the approved architecture while meeting the requirement that new sites must have consistent repeatable baseline controls. Staged delivery, attributable change evidence and rollback guard against an otherwise avoidable service disruption.
Incorrect Answers
A:
Manual one-off changes are hard to reproduce or audit. For this question, it is not a defensible choice for controlled implementation and it does not satisfy the requirement that new sites must have consistent repeatable baseline controls.
B:
The alternative “Publish reusable Terraform modules with policy tests and documented ownership” describes the chosen design, but not how it is introduced safely. In the reported observation, asset review shows inconsistent IAM, patch settings and logging between regions. The option “Migrate one region to versioned modules and controlled state storage” instead addresses the requested controlled implementation.
D:
The alternative “Monitor module drift, change approvals and obsolete versions” addresses later monitoring rather than the actual controlled deployment. In the reported observation, asset review shows inconsistent IAM, patch settings and logging between regions. The option “Migrate one region to versioned modules and controlled state storage” instead addresses the requested controlled implementation.
Question 979
At Regional financial adviser, the private research assistant workload has a documented risk: employees want document-grounded answers without granting a chatbot access to privileged trading APIs. The review found that demo agent follows instructions embedded in retrieved research PDFs. The architecture must also ensure that approved internal reports must be cited and trading authorization separate. Select TWO complementary actions: one durable target architecture and one controlled way to introduce it while preserving service. Which TWO are best?
- Map trusted retrieved context and the tool-authorization boundary
- Watch tool-denial logs, citation fidelity and document-access grants
- Introduce approved corpus sources with prompt-injection tests and read-only permissions
- Ground answers in vetted enterprise search and isolate actions behind explicit identity checks
- Treat every retrieved document as an instruction from the security administrator
Explanation
Correct Answers
C:
Introduce approved corpus sources with prompt-injection tests and read-only permissions is a controlled means to introduce the approved architecture while meeting the requirement that approved internal reports must be cited and trading authorization separate. 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.
D:
Ground answers in vetted enterprise search and isolate actions behind explicit identity checks targets the underlying issue (employees want document-grounded answers without granting a chatbot access to privileged trading APIs) rather than merely reporting symptoms. It also preserves the required outcome (approved internal reports must be cited and trading authorization separate), which a target-state design must satisfy. This supplies the lasting-control half of the two-action package.
Incorrect Answers
A:
This diagnoses the original issue (employees want document-grounded answers without granting a chatbot access to privileged trading APIs), but the TWO requested actions are the target architecture and its implementation; discovery was already completed.
B:
Monitoring private research assistant for future drift is useful after the rollout, but it does not specify the requested target control and safe implementation for this case: employees want document-grounded answers without granting a chatbot access to privileged trading APIs.
E:
Untrusted document content cannot grant privileged tool authority. The shortcut does not supply the requested design and safe rollout for private research assistant, where demo agent follows instructions embedded in retrieved research PDFs.
Question 980
At Manufacturer predictive maintenance team, the Agent Platform ML pipelines workload has a documented risk: each retraining run uses a different preprocessing notebook with no lineage. The review found that model results diverge even when training data IDs match. The architecture must also ensure that factory anomaly models need reproducible releases and operator approval. What follow-up metric or alert will identify emerging failures after the initial acceptance test?
- Use versioned ML pipelines with recorded artifacts and reproducible evaluation gates
- Train directly on mutable shared notebooks without tracking artifact versions
- Monitor pipeline failures, model drift and promotion exceptions
- Re-run the same dataset and verify matching approved artifact lineage
Explanation
Correct Answer
C:
Monitor pipeline failures, model drift and promotion exceptions reveals recurrence, drift or changing dependencies after launch. It provides a repeatable signal for the risk (each retraining run uses a different preprocessing notebook with no lineage) rather than treating the first successful test as permanent assurance.
Incorrect Answers
A:
The alternative “Use versioned ML pipelines with recorded artifacts and reproducible evaluation gates” selects a design, but does not provide the ongoing signal required to detect drift. In the reported observation, model results diverge even when training data IDs match. The option “Monitor pipeline failures, model drift and promotion exceptions” instead addresses the requested ongoing operational assurance.
B:
An unversioned notebook cannot support reproducible evaluation. For this question, it is not a defensible choice for ongoing operational assurance and it does not satisfy the requirement that factory anomaly models need reproducible releases and operator approval.
D:
The alternative “Re-run the same dataset and verify matching approved artifact lineage” is a one-time acceptance result rather than ongoing operational assurance. In the reported observation, model results diverge even when training data IDs match. The option “Monitor pipeline failures, model drift and promotion exceptions” instead addresses the requested ongoing operational assurance.