100% Latest & Updated Google Professional Cloud Architect Practice Test Questions, Exam Dumps & Verified Answers!
30 Days Free Updates, Instant Download!
Question 300
A loyalty API has multiple zones but no one has tested how it behaves when a zone or backing cache disappears. Which operational action best maintains reliability feedback loop after launch?
- File the game-day report without fixing identified flaws
- Turn game-day findings into tracked fixes and rerun the specific failed experiment
- Repeat uncontrolled experiments until an outage occurs
- Assume future deployments cannot change dependencies
Explanation
Correct Answer
B:Chaos exercises create value only when they change resilience defects
Incorrect Answers
A:Does not improve reliability. Chaos exercises create value only when they change resilience defects.
C:Not a managed learning cycle. Chaos exercises create value only when they change resilience defects.
D:Resilience can drift. Chaos exercises create value only when they change resilience defects.
Question 301
A company exposes a financial application through a new API gateway and wants to validate abuse resistance before launch. Which architecture most directly meets the need for risk-based security verification?
- Conduct authorized penetration testing focused on gateway authentication and authorization boundaries
- Use a broad unauthenticated API to simplify external integration
- Replace authorization testing with a network scan only
- Treat TLS alone as sufficient client authorization
Explanation
Correct Answer
A:Security testing should exercise likely abuse paths rather than rely only on network scanning
Incorrect Answers
B:Violates financial access needs. Security testing should exercise likely abuse paths rather than rely only on network scanning.
C:Object-level checks are missing. Security testing should exercise likely abuse paths rather than rely only on network scanning.
D:Transport privacy is not access control. Security testing should exercise likely abuse paths rather than rely only on network scanning.
Question 302
A company exposes a financial application through a new API gateway and wants to validate abuse resistance before launch. Which implementation choice most directly delivers safe penetration testing scope?
- Document target systems, permitted techniques and rollback contacts before running tests
- Launch penetration tests without agreed target scope
- Probe customer production data without authorized limits
- Skip rollback contacts because a tester has experience
Explanation
Correct Answer
A:Authorized testing requires containment to avoid harming unrelated tenant services
Incorrect Answers
B:May harm unrelated services. Authorized testing requires containment to avoid harming unrelated tenant services.
C:Unacceptable testing risk. Authorized testing requires containment to avoid harming unrelated tenant services.
D:No safe escalation route. Authorized testing requires containment to avoid harming unrelated tenant services.
Question 303
A company exposes a financial application through a new API gateway and wants to validate abuse resistance before launch. Which validation would best establish proof broken access is rejected?
- Test only incorrect passwords against the login endpoint
- Check API gateway uptime
- Attempt cross-tenant object access with a valid but unauthorized API identity
- Scan open ports without inspecting object-level policies
Explanation
Correct Answer
C:A test with an invalid password does not demonstrate object-level authorization
Incorrect Answers
A:Misses valid-user cross-tenant access. A test with an invalid password does not demonstrate object-level authorization.
B:Not authorization denial. A test with an invalid password does not demonstrate object-level authorization.
D:Can miss broken access control. A test with an invalid password does not demonstrate object-level authorization.
Question 304
A company exposes a financial application through a new API gateway and wants to validate abuse resistance before launch. Which operational action best maintains ongoing security validation after launch?
- Mark vulnerabilities closed before retesting fixed endpoints
- Run security tests only once before product launch
- Track only the number of findings instead of exploitable severity
- Retest high-severity fixes and include regression cases in the release pipeline
Explanation
Correct Answer
D:One prelaunch penetration test cannot cover later gateway and permission changes
Incorrect Answers
A:Not remediation evidence. One prelaunch penetration test cannot cover later gateway and permission changes.
B:New releases may reopen flaws. One prelaunch penetration test cannot cover later gateway and permission changes.
C:Not control effectiveness. One prelaunch penetration test cannot cover later gateway and permission changes.
Question 305
A streaming service wants to validate availability under sudden traffic growth and deliberate loss of a streaming region. Which architecture most directly meets the need for capacity and fault tolerance test strategy?
- Combine representative load testing with controlled regional-failure scenarios
- Run only a throughput benchmark with no region-loss simulation
- Use a region failure exercise with no realistic user load
- Review infrastructure design instead of executing tests
Explanation
Correct Answer
A:High load and regional disruption stress different aspects of reliability
Incorrect Answers
B:Does not test failure tolerance. High load and regional disruption stress different aspects of reliability.
C:May understate capacity issues. High load and regional disruption stress different aspects of reliability.
D:Not observed resilience. High load and regional disruption stress different aspects of reliability.
Question 306
A streaming service wants to validate availability under sudden traffic growth and deliberate loss of a streaming region. Which implementation choice most directly delivers bounded load-exercise controls?
- Define stop thresholds and reserve realistic test traffic capacity before the exercise
- Launch full-scale synthetic traffic without resource stop thresholds
- Ignore other tenants on shared quota during testing
- Change load scripts during results comparison with no record
Explanation
Correct Answer
A:A load test must not unexpectedly exhaust critical production shared quotas
Incorrect Answers
B:Can exhaust production. A load test must not unexpectedly exhaust critical production shared quotas.
C:Risks collateral effects. A load test must not unexpectedly exhaust critical production shared quotas.
D:Makes evidence hard to interpret. A load test must not unexpectedly exhaust critical production shared quotas.
Question 307
A streaming service wants to validate availability under sudden traffic growth and deliberate loss of a streaming region. Which validation would best establish proof business video playback remains usable?
- Measure only backend HTTP throughput and ignore playback quality
- Check CPU averages rather than time-to-recover
- Count issued stream URLs instead of playable sessions
- Measure start failures, rebuffering and recovery time throughout fault injection
Explanation
Correct Answer
D:Healthy backend CPU usage does not prove viewers can watch streams
Incorrect Answers
A:Users can still see rebuffers. Healthy backend CPU usage does not prove viewers can watch streams.
B:Misses service degradation. Healthy backend CPU usage does not prove viewers can watch streams.
C:Not end-user success. Healthy backend CPU usage does not prove viewers can watch streams.
Question 308
A streaming service wants to validate availability under sudden traffic growth and deliberate loss of a streaming region. Which operational action best maintains follow-up performance improvement after launch?
- Increase traffic blindly whenever rebuffering worsens
- Archive results without tuning bottlenecked subsystems
- Keep same experiment despite discovered unsafe limits
- Use observed bottlenecks to adjust autoscaling and region routing before the next release
Explanation
Correct Answer
D:A report without engineering changes does not improve streaming resilience
Incorrect Answers
A:Can amplify incident. A report without engineering changes does not improve streaming resilience.
B:No reliability improvement. A report without engineering changes does not improve streaming resilience.
C:Repeat risk without learning. A report without engineering changes does not improve streaming resilience.
Question 309
A video platform rolls out a new container release and needs to keep the existing version available for immediate rollback. Which architecture most directly meets the need for low-risk progressive service delivery?
- Deploy all users to untested new revision at once
- Use Cloud Deploy or supported traffic splitting to promote tested container revisions progressively
- Keep only the new build artifact and discard stable rollback revisions
- Publish code changes without linking them to production revision identity
Explanation
Correct Answer
B:Phased traffic limits customer impact while retaining a rollback target
Incorrect Answers
A:Maximizes blast radius. Phased traffic limits customer impact while retaining a rollback target.
C:Removes recovery path. Phased traffic limits customer impact while retaining a rollback target.
D:Not controlled promotion. Phased traffic limits customer impact while retaining a rollback target.
Question 310
A video platform rolls out a new container release and needs to keep the existing version available for immediate rollback. Which implementation choice most directly delivers release readiness gates?
- Promote builds based only on successful compilation
- Bind deployment promotion to integration tests and visible change approval
- Allow any developer identity to bypass approvals without logging
- Use mutable latest tags as the only release record
Explanation
Correct Answer
B:Build completion alone does not establish a new revision is safe for customers
Incorrect Answers
A:Misses integration/runtime regressions. Build completion alone does not establish a new revision is safe for customers.
C:Undermines release governance. Build completion alone does not establish a new revision is safe for customers.
D:Cannot reliably identify approved bytes. Build completion alone does not establish a new revision is safe for customers.
Google Cloud Certified - Professional Cloud Architect (Professional Cloud Architect) Practice Tests