100% Latest & Updated Google Professional Cloud Architect Practice Test Questions, Exam Dumps & Verified Answers!
30 Days Free Updates, Instant Download!
Question 30
A public image catalog serves millions of geographically dispersed viewers and origin latency drives poor page responsiveness. Which implementation choice most directly delivers cache-correct publishing?
- Use the same mutable URL for every version without a cache policy
- Set content cache keys and explicit cache-control headers for versioned media
- Invalidate every object every minute
- Set every public image response to no-store
Explanation
Correct Answer
B:Well-defined cache behavior avoids serving stale images after asset updates
Incorrect Answers
A:May keep old content in client and edge caches. Well-defined cache behavior avoids serving stale images after asset updates.
C:Destroys caching efficiency and increases origin demand. Well-defined cache behavior avoids serving stale images after asset updates.
D:Turns off the edge-cache benefit. Well-defined cache behavior avoids serving stale images after asset updates.
Question 31
A public image catalog serves millions of geographically dispersed viewers and origin latency drives poor page responsiveness. Which validation would best establish validation of geographic latency improvement?
- Compare regional client p95 response times and cache-hit metrics under realistic load
- Benchmark a client co-located with the origin only
- Count cached objects without measuring user response time
- Compare download price for two object storage classes
Explanation
Correct Answer
A:User-region timings and cache evidence demonstrate benefit rather than local-origin tests
Incorrect Answers
B:Ignores geographically distant viewers. User-region timings and cache evidence demonstrate benefit rather than local-origin tests.
C:Cache inventory alone does not show performance improvement. User-region timings and cache evidence demonstrate benefit rather than local-origin tests.
D:Does not prove latency benefits. User-region timings and cache evidence demonstrate benefit rather than local-origin tests.
Question 32
A public image catalog serves millions of geographically dispersed viewers and origin latency drives poor page responsiveness. Which operational action best maintains cache quality during new releases after launch?
- Purge the entire cache after every single view
- Watch only origin disk capacity
- Track cache hit ratio and selectively invalidate changed content when necessary
- Turn off CDN on deployment days without testing changed asset versions
Explanation
Correct Answer
C:Cache observability limits unnecessary origin load and controls stale content
Incorrect Answers
A:Wastes cached delivery and increases backend work. Cache observability limits unnecessary origin load and controls stale content.
B:Does not monitor cache freshness or regional hit rate. Cache observability limits unnecessary origin load and controls stale content.
D:Causes unnecessary cache misses and latency. Cache observability limits unnecessary origin load and controls stale content.
Question 33
A containerized forecasting API is idle at night but spikes each morning; cold starts now breach its p99 response target. Which architecture most directly meets the need for fast response with elastic capacity?
- Set minimum instances to zero for a known latency-sensitive idle-to-burst workload
- Allocate peak instance count permanently with no measurements
- Keep a measured minimum of warm Cloud Run instances while retaining autoscaling
- Move the stateless API into a scheduled nightly batch job
Explanation
Correct Answer
C:Minimum instances reduce startup delay without requiring all peak capacity all day
Incorrect Answers
A:Permits the same cold-start problem. Minimum instances reduce startup delay without requiring all peak capacity all day.
B:Spends on idle capacity beyond the observed need. Minimum instances reduce startup delay without requiring all peak capacity all day.
D:Cannot serve interactive morning requests. Minimum instances reduce startup delay without requiring all peak capacity all day.
Question 34
A containerized forecasting API is idle at night but spikes each morning; cold starts now breach its p99 response target. Which implementation choice most directly delivers latency-sensitive rollout controls?
- Configure measured minimum instances and concurrency per revision
- Change only request timeout without keeping any warm capacity
- Set concurrency to its maximum without checking p99 latency
- Disable revision level scaling controls without a performance test
Explanation
Correct Answer
A:Warm capacity and request concurrency shape cold-start exposure under bursts
Incorrect Answers
B:Does not reduce first-instance startup latency. Warm capacity and request concurrency shape cold-start exposure under bursts.
C:Overloaded per-instance processing may worsen tail latency. Warm capacity and request concurrency shape cold-start exposure under bursts.
D:Removes safeguards without demonstrating fewer cold starts. Warm capacity and request concurrency shape cold-start exposure under bursts.
Question 35
A containerized forecasting API is idle at night but spikes each morning; cold starts now breach its p99 response target. Which validation would best establish proof that the p99 target is achieved?
- Test only requests after the service has been warm for hours
- Measure first-request and burst p99 latency after an idle interval
- Calculate mean latency without p99 burst measurements
- Measure just the number of available container images
Explanation
Correct Answer
B:Idle-to-burst probes reproduce the actual cold-start failure mode
Incorrect Answers
A:Omits the initial cold-start spike. Idle-to-burst probes reproduce the actual cold-start failure mode.
C:Tail latency can fail while the average looks healthy. Idle-to-burst probes reproduce the actual cold-start failure mode.
D:Does not exercise request startup behavior. Idle-to-burst probes reproduce the actual cold-start failure mode.
Question 36
A containerized forecasting API is idle at night but spikes each morning; cold starts now breach its p99 response target. Which operational action best maintains capacity and cost rebalancing after launch?
- Keep full peak minimum capacity through every overnight idle period
- Optimize only the average CPU metric
- Tune minimum instances from p99 latency and idle cost measurements
- Raise the response SLO without agreement from application owners
Explanation
Correct Answer
C:A fixed large minimum wastes capacity while an undersized one breaches latency
Incorrect Answers
A:Can negate the economic benefit of autoscaling. A fixed large minimum wastes capacity while an undersized one breaches latency.
B:May miss p99 cold-start response effects. A fixed large minimum wastes capacity while an undersized one breaches latency.
D:Changes the requirement rather than solving it. A fixed large minimum wastes capacity while an undersized one breaches latency.
Question 37
A batch geospatial pipeline processes new maps every night under a completion deadline but has no interactive users. Which architecture most directly meets the need for deadline-aware ephemeral batch capacity?
- Use a fixed high-cost always-on web fleet for nightly batch tasks
- Serve the job through a latency-optimized user-facing load balancer
- Run a single manual VM that an operator starts when convenient
- Use Cloud Batch with suitable Compute Engine task resources and job parallelism
Explanation
Correct Answer
D:Managed batch scheduling fits finite noninteractive work with explicit completion requirements
Incorrect Answers
A:Pays for resources outside the finite processing window. Managed batch scheduling fits finite noninteractive work with explicit completion requirements.
B:Interactive routing does not provide deadline-aware batch scheduling. Managed batch scheduling fits finite noninteractive work with explicit completion requirements.
C:Adds unpredictable startup and failure risk. Managed batch scheduling fits finite noninteractive work with explicit completion requirements.
Question 38
A batch geospatial pipeline processes new maps every night under a completion deadline but has no interactive users. Which implementation choice most directly delivers controllable capacity for night jobs?
- Set task retry and resource limits and choose suitable provisioning options for deadline criticality
- Place retry state solely on local ephemeral VM disk
- Provision minimum capacity without considering full job deadline
- Set unbounded retries with no completion monitoring
Explanation
Correct Answer
A:Retries and compute sizing must account for interruptions and available runtime
Incorrect Answers
B:Preemption can erase useful progress. Retries and compute sizing must account for interruptions and available runtime.
C:Insufficient parallelism may miss the cutoff. Retries and compute sizing must account for interruptions and available runtime.
D:Failed tasks can silently consume the schedule. Retries and compute sizing must account for interruptions and available runtime.
Question 39
A batch geospatial pipeline processes new maps every night under a completion deadline but has no interactive users. Which validation would best establish validation of a batch service objective?
- Confirm that the scheduler accepted the job
- Benchmark one tiny sample and extrapolate without validation
- Run full-size sample jobs and measure completion time and retry recovery
- Check only aggregate VM CPU consumption
Explanation
Correct Answer
C:A job-start notification does not establish deadline compliance
Incorrect Answers
A:Acceptance is not final artifact delivery. A job-start notification does not establish deadline compliance.
B:Does not show full-size performance. A job-start notification does not establish deadline compliance.
D:Does not measure completion time or recovery. A job-start notification does not establish deadline compliance.
Question 40
A batch geospatial pipeline processes new maps every night under a completion deadline but has no interactive users. Which operational action best maintains late-job remediation after launch?
- Monitor costs once a month after reports fail
- Retry the entire job from the start whenever one task fails
- Alert on projected completion beyond the deadline and adjust parallelism or capacity
- Keep a large fixed fleet online all day
Explanation
Correct Answer
C:Deadline-aware signals support capacity interventions before business users miss a map refresh
Incorrect Answers
A:Misses same-night deadline risk. Deadline-aware signals support capacity interventions before business users miss a map refresh.
B:Wastes completed work and may miss the cutoff. Deadline-aware signals support capacity interventions before business users miss a map refresh.
D:Violates the batch cost constraint. Deadline-aware signals support capacity interventions before business users miss a map refresh.
Google Cloud Certified - Professional Cloud Architect (Professional Cloud Architect) Practice Tests