Google Professional Cloud Security Engineer Practice Test 71
Question 700
An application requires key usage to stop when an external approval service revokes the key. An engineer must explain the recommendation for External Key Manager custody. Which underlying principle is the MOST accurate?
- Encryption in transit needs endpoint identity validation in addition to encrypted bytes.
- Cloud EKM shifts external key custody and introduces dependency on external key-management availability.
- Storage lifecycle rules automate object state changes under conditions separate from IAM authorization.
- Minimization and lifecycle governance reduce residual sensitive-data exposure beyond IAM encryption.
Explanation
Correct Answer
B:Cloud EKM shifts external key custody and introduces dependency on external key-management availability. Therefore the implementation decision is to use supported Cloud External Key Manager integration and design for external availability and permission boundaries.
Incorrect Answers
A:Unavailable external keys can make protected cloud data inaccessible even when IAM allows access. Encryption in transit needs endpoint identity validation in addition to encrypted bytes.
C:Unavailable external keys can make protected cloud data inaccessible even when IAM allows access. Storage lifecycle rules automate object state changes under conditions separate from IAM authorization.
D:Unavailable external keys can make protected cloud data inaccessible even when IAM allows access. This describes the separate Sensitive data retention and data lifecycle mechanism; the present case instead requires External Key Manager custody.
Question 701
An agent uses connectors and enterprise tools across multiple governed data sources. An engineer must explain the recommendation for Gemini Enterprise Agent Platform tool permission design. Which underlying principle is the MOST accurate?
- AI runtime threat detection requires operational telemetry and governed diagnostic retention.
- Metadata server tokens reflect attached workload identity and must be protected from untrusted code paths.
- IaaS-hosted models leave more guest and runtime controls with customers than managed AI services.
- Agent tool execution authority must be narrower than the natural-language instruction itself.
Explanation
Correct Answer
D:Agent tool execution authority must be narrower than the natural-language instruction itself. Therefore the implementation decision is to configure least-privilege agent identity and connector permissions with explicit action approval gates.
Incorrect Answers
A:AI runtime threat detection requires operational telemetry and governed diagnostic retention. This describes the separate AI model monitoring and abuse investigations mechanism; the present case instead requires Gemini Enterprise Agent Platform tool permission design. Allowing an agent's conversational confidence to authorize high-impact writes bypasses controls.
B:Metadata server tokens reflect attached workload identity and must be protected from untrusted code paths. This describes the separate Compute metadata exposure hardening mechanism; the present case instead requires Gemini Enterprise Agent Platform tool permission design. Allowing an agent's conversational confidence to authorize high-impact writes bypasses controls.
C:IaaS-hosted models leave more guest and runtime controls with customers than managed AI services. This describes the separate IaaS versus managed AI training boundary mechanism; the present case instead requires Gemini Enterprise Agent Platform tool permission design. Allowing an agent's conversational confidence to authorize high-impact writes bypasses controls.
Question 702
A Terraform change would allow inbound SSH from every public address. Which approach BEST meets the requirement while respecting the system constraints?
- Scan infrastructure-as-code plans against approved security guardrails and require review before deployment.
- Trust every Terraform apply simply because its syntax is valid.
- Enable Binary Authorization enforcement on GKE with an appropriate signed attestation policy.
- Enable applicable Cloud Run Binary Authorization with reviewed service and job policy semantics.
Explanation
Correct Answer
A:Scan infrastructure-as-code plans against approved security guardrails and require review before deployment. IaC policy validation can catch configuration drift earlier than production vulnerability alerts. Verification should demonstrate that an intentionally unsafe plan fails the gate while a least-privilege revision passes.
Incorrect Answers
B:Trust every Terraform apply simply because its syntax is valid. This is an unsafe or ineffective shortcut for Infrastructure as code security validation. A review comment without enforced CI policy can still permit unsafe infrastructure rollout.
C:Enable Binary Authorization enforcement on GKE with an appropriate signed attestation policy. This primarily addresses Binary Authorization GKE admission, not the specific Infrastructure as code security validation decision. Here, iaC policy validation can catch configuration drift earlier than production vulnerability alerts.
D:Enable applicable Cloud Run Binary Authorization with reviewed service and job policy semantics. This primarily addresses Binary Authorization Cloud Run revision policy, not the specific Infrastructure as code security validation decision. Here, iaC policy validation can catch configuration drift earlier than production vulnerability alerts.
Question 703
A load balancer forwards to a backend that requires encrypted application traffic end to end. Testing exposes this failure: Terminating TLS at the edge does not automatically encrypt every load balancer-to-backend hop. Which remediation should be applied FIRST?
- Assume any HTTP health check proves all downstream customer data is encrypted end to end.
- Design a supported Certificate Authority Service CA pool and narrowly delegated certificate issuance.
- Enable DNSSEC on the supported Cloud DNS public zone and publish matching trust-chain records.
- Inspect certificate lifecycle, forwarding rule, backend service and health check settings.
Explanation
Correct Answer
D:Inspect certificate lifecycle, forwarding rule, backend service and health check settings. Confirmation requires seeing that certificate validation and backend health checks succeed without sending protected data over unintended plaintext paths.
Incorrect Answers
A:Assume any HTTP health check proves all downstream customer data is encrypted end to end. This retains the HTTPS load balancer certificate and backend TLS problem. Frontend TLS and backend TLS are separate transport protection design choices.
B:Design a supported Certificate Authority Service CA pool and narrowly delegated certificate issuance. It addresses Certificate Authority Service private PKI, not the diagnosed HTTPS load balancer certificate and backend TLS failure. Terminating TLS at the edge does not automatically encrypt every load balancer-to-backend hop.
C:Enable DNSSEC on the supported Cloud DNS public zone and publish matching trust-chain records. It addresses Cloud DNSSEC authoritative zone protection, not the diagnosed HTTPS load balancer certificate and backend TLS failure. Terminating TLS at the edge does not automatically encrypt every load balancer-to-backend hop.
Question 704
An incident requires retiring a compromised cryptographic key while preserving critical data recovery. After the recommended KMS key version destruction and restore window change is implemented, which observation MOST directly verifies that it works?
- The data service can encrypt and decrypt using the intended key while unrelated principals cannot.
- Incidents correlate sensitive operations to request identities without exposing raw private prompts unnecessarily.
- Both sides authenticate the expected certificate chain and unauthorized service identities are denied.
- Data access behavior and any permitted restore window are understood before irreversible destruction.
Explanation
Correct Answer
D:Data access behavior and any permitted restore window are understood before irreversible destruction. This directly tests whether the KMS key version destruction and restore window goal was met. Cryptographic key destruction can make encrypted data unrecoverable regardless of object retention.
Incorrect Answers
A:The data service can encrypt and decrypt using the intended key while unrelated principals cannot. This is evidence about Cloud KMS CMEK service agent access, not direct proof of the required KMS key version destruction and restore window behavior. The relevant check is that data access behavior and any permitted restore window are understood before irreversible destruction.
B:Incidents correlate sensitive operations to request identities without exposing raw private prompts unnecessarily. This is evidence about AI model monitoring and abuse investigations, not direct proof of the required KMS key version destruction and restore window behavior. The relevant check is that data access behavior and any permitted restore window are understood before irreversible destruction.
C:Both sides authenticate the expected certificate chain and unauthorized service identities are denied. This is evidence about Mutual TLS for service-to-service identity, not direct proof of the required KMS key version destruction and restore window behavior. The relevant check is that data access behavior and any permitted restore window are understood before irreversible destruction.
Question 705
A team uses per-object ACLs inconsistently and wants centralized bucket authorization. Which approach BEST meets the requirement while respecting the system constraints?
- Split deployment, access-policy editing and security evidence administration among distinct identities.
- Enable uniform bucket-level access where appropriate and manage object access through IAM.
- Assume changing a bucket label instantly deletes all legacy object ACL semantics without enabling uniform access.
- Define and test an Access Context Manager access level with intended device and network conditions.
Explanation
Correct Answer
B:Enable uniform bucket-level access where appropriate and manage object access through IAM. Uniform bucket-level access disables object ACL evaluation in favor of IAM policies. Verification should demonstrate that aCL-based grants no longer provide unintended access and IAM-defined users continue to work.
Incorrect Answers
A:Split deployment, access-policy editing and security evidence administration among distinct identities. This primarily addresses IAM separation of security duties, not the specific Cloud Storage ACL versus IAM with uniform access decision. Here, uniform bucket-level access disables object ACL evaluation in favor of IAM policies.
C:Assume changing a bucket label instantly deletes all legacy object ACL semantics without enabling uniform access. This is an unsafe or ineffective shortcut for Cloud Storage ACL versus IAM with uniform access. Retaining object ACL assumptions can leave applications unexpectedly denied after migration.
D:Define and test an Access Context Manager access level with intended device and network conditions. This primarily addresses Access Context Manager access levels, not the specific Cloud Storage ACL versus IAM with uniform access decision. Here, uniform bucket-level access disables object ACL evaluation in favor of IAM policies.
Question 706
A sensitive administration workflow should be available only from managed devices or approved corporate IP ranges. After the recommended Access Context Manager access levels change is implemented, which observation MOST directly verifies that it works?
- The identified rule explains successful and failed operations for the exact principal and resource.
- Requests that satisfy the access level are eligible while unmatched contexts are rejected by the consuming policy.
- The leaked key can no longer authenticate and supported workloads still operate using their new identity.
- Onboarding and removal change effective access in the intended projects without unrelated grants.
Explanation
Correct Answer
B:Requests that satisfy the access level are eligible while unmatched contexts are rejected by the consuming policy. This directly tests whether the Access Context Manager access levels goal was met. Access levels describe contextual trust conditions and must be attached to an enforcing service policy.
Incorrect Answers
A:The identified rule explains successful and failed operations for the exact principal and resource. This is evidence about IAM effective permission troubleshooting, not direct proof of the required Access Context Manager access levels behavior. The relevant check is that requests that satisfy the access level are eligible while unmatched contexts are rejected by the consuming policy.
C:The leaked key can no longer authenticate and supported workloads still operate using their new identity. This is evidence about Service account key inventory and exposure response, not direct proof of the required Access Context Manager access levels behavior. The relevant check is that requests that satisfy the access level are eligible while unmatched contexts are rejected by the consuming policy.
D:Onboarding and removal change effective access in the intended projects without unrelated grants. This is evidence about Group-based IAM authorization, not direct proof of the required Access Context Manager access levels behavior. The relevant check is that requests that satisfy the access level are eligible while unmatched contexts are rejected by the consuming policy.
Question 707
A permitted principal inside a service perimeter receives a BigQuery permission denied error. Which approach BEST meets the requirement while respecting the system constraints?
- Add narrowly scoped VPC Service Controls ingress or egress rules using principal and service operation constraints.
- Disable all Cloud Storage IAM checks because a VPC Service Controls perimeter is enabled.
- Apply VPC Service Controls perimeters to supported data services with least-privilege ingress and egress policies.
- Resolve IAM authorization and service perimeter evaluation as separate controls.
Explanation
Correct Answer
D:Resolve IAM authorization and service perimeter evaluation as separate controls. VPC Service Controls boundaries do not replace IAM roles on protected data resources. Verification should demonstrate that tests show the caller has both sufficient resource permissions and a permitted perimeter path.
Incorrect Answers
A:Add narrowly scoped VPC Service Controls ingress or egress rules using principal and service operation constraints. This primarily addresses VPC Service Controls ingress egress policy, not the specific VPC Service Controls versus IAM authorization decision. Here, vPC Service Controls boundaries do not replace IAM roles on protected data resources.
B:Disable all Cloud Storage IAM checks because a VPC Service Controls perimeter is enabled. This is an unsafe or ineffective shortcut for VPC Service Controls versus IAM authorization. Treating perimeter membership as an automatic IAM grant can expose false assumptions about access.
C:Apply VPC Service Controls perimeters to supported data services with least-privilege ingress and egress policies. This primarily addresses VPC Service Controls service perimeter, not the specific VPC Service Controls versus IAM authorization decision. Here, vPC Service Controls boundaries do not replace IAM roles on protected data resources.
Question 708
A data team stores customer support transcripts that may contain government identifiers. After the recommended Sensitive Data Protection PII inspection change is implemented, which observation MOST directly verifies that it works?
- Deletion before retention expiry is rejected and protected retention settings cannot be casually shortened.
- The same query yields permitted rows for each test identity and denies or filters restricted rows.
- Findings identify sensitive matches with record-level access governed by project policy.
- Authorized bucket principals retain access and object-specific ACL exceptions stop applying.
Explanation
Correct Answer
C:Findings identify sensitive matches with record-level access governed by project policy. This directly tests whether the Sensitive Data Protection PII inspection goal was met. Sensitive Data Protection classifies sensitive content but does not automatically revoke every reader's IAM permissions.
Incorrect Answers
A:Deletion before retention expiry is rejected and protected retention settings cannot be casually shortened. This is evidence about Cloud Storage retention policy and Bucket Lock, not direct proof of the required Sensitive Data Protection PII inspection behavior. The relevant check is that findings identify sensitive matches with record-level access governed by project policy.
B:The same query yields permitted rows for each test identity and denies or filters restricted rows. This is evidence about BigQuery row-level access policy, not direct proof of the required Sensitive Data Protection PII inspection behavior. The relevant check is that findings identify sensitive matches with record-level access governed by project policy.
D:Authorized bucket principals retain access and object-specific ACL exceptions stop applying. This is evidence about Cloud Storage uniform bucket-level permissions, not direct proof of the required Sensitive Data Protection PII inspection behavior. The relevant check is that findings identify sensitive matches with record-level access governed by project policy.
Question 709
VMs in a private subnet need to download package updates from public repositories without public IPs. Which approach BEST meets the requirement while respecting the system constraints?
- Configure Cloud Router BGP peers and appropriate custom or default route advertisements.
- Assume adding Cloud NAT makes every internal VM directly reachable from the public internet.
- Create supported Private Service Connect endpoint connectivity for Google APIs with matching DNS.
- Configure Cloud NAT with appropriate gateway, addresses and subnet egress coverage.
Explanation
Correct Answer
D:Configure Cloud NAT with appropriate gateway, addresses and subnet egress coverage. Cloud NAT translates eligible outbound connections and is not an inbound proxy or firewall substitute. Verification should demonstrate that private VMs establish allowed outbound connections and unsolicited inbound internet traffic remains unavailable.
Incorrect Answers
A:Configure Cloud Router BGP peers and appropriate custom or default route advertisements. This primarily addresses Cloud Router dynamic BGP advertisement, not the specific Cloud NAT outbound without inbound exposure decision. Here, cloud NAT translates eligible outbound connections and is not an inbound proxy or firewall substitute.
B:Assume adding Cloud NAT makes every internal VM directly reachable from the public internet. This is an unsafe or ineffective shortcut for Cloud NAT outbound without inbound exposure. Treating Cloud NAT as inbound protection while allowing broad alternative ingress can leave exposure.
C:Create supported Private Service Connect endpoint connectivity for Google APIs with matching DNS. This primarily addresses Private Service Connect Google APIs endpoints, not the specific Cloud NAT outbound without inbound exposure decision. Here, cloud NAT translates eligible outbound connections and is not an inbound proxy or firewall substitute.
Question 710
An automation service maintains a security group for a rotating set of approved analysts. If the relevant Programmatic group membership governance safeguard is omitted, which outcome is the MOST directly associated failure?
- Allowing any deployer to attach an owner-level service account can enable privilege escalation.
- Unverified group automation may silently grant inherited IAM roles to unintended principals.
- Routine use of shared super-admin credentials destroys accountability and enlarges blast radius.
- Checking only successful SSO sessions overlooks excess permissions inherited through groups.
Explanation
Correct Answer
B:Unverified group automation may silently grant inherited IAM roles to unintended principals. Managing authorization through groups can simplify changes but group administration itself is privileged. The corrective approach is to reconcile approved roster changes against group membership and resulting project access.
Incorrect Answers
A:This failure pattern belongs chiefly to Scoped permission for service account creation; it is not the failure most directly tested in the Programmatic group membership governance situation.
C:Routine use of shared super-admin credentials destroys accountability and enlarges blast radius. This failure pattern belongs chiefly to Super administrator emergency access; it is not the failure most directly tested in the Programmatic group membership governance situation. Unverified group automation may silently grant inherited IAM roles to unintended principals.
D:Checking only successful SSO sessions overlooks excess permissions inherited through groups. This failure pattern belongs chiefly to Group-based provisioning audit trail; it is not the failure most directly tested in the Programmatic group membership governance situation. Unverified group automation may silently grant inherited IAM roles to unintended principals.