Cost Optimization

Billing, cost management, and resource optimization

1

Set 1

20 questions • ~25 min

Start Set 1
2

Set 2

20 questions • ~25 min

Start Set 2
3

Set 3

20 questions • ~25 min

Start Set 3
4

Set 4

20 questions • ~25 min

Start Set 4
5

Set 5

20 questions • ~25 min

Start Set 5
6

Set 6

20 questions • ~25 min

Start Set 6
7

Set 7

4 questions • ~25 min

Start Set 7

All Questions

124 questions • Flexible time

Start All

Quick Drill

15 questions • ~15 min

Quick Drill
Choose Different Domain

Sample GCP PCA questions with explanations

These six general GCP PCA questions come from the full practice bank; they are not filtered to the selected domain. Review the answers and explanations for cross-domain practice, then choose a focused set above.

Question 1 General Choose 2

You're optimizing a Dataflow pipeline that processes 1TB of data daily. Which TWO approaches can improve performance?

  • ✓ Adjust worker machine types and number of workers for parallelization.
  • ✓ Use efficient data formats (Avro, Parquet) instead of JSON/CSV.
  • ○ Process data sequentially on one worker.
  • ○ Increase shuffle file size.
  • ○ Disable autoscaling.
Answer: A. Adjust worker machine types and number of workers for parallelization. | B. Use efficient data formats (Avro, Parquet) instead of JSON/CSV.
Right-sizing workers and using efficient formats reduce processing time. Sequential processing is slow, and disabling autoscaling reduces flexibility.
Question 2 General

You want to set up an uptime check that monitors your API endpoint from multiple global locations and alerts if the endpoint is down. What should you configure?

  • ✓ Cloud Monitoring Uptime Checks with multiple regions selected.
  • ○ Cron job pinging the endpoint.
  • ○ Cloud Scheduler calling the API.
  • ○ Manual checks from different locations.
Answer: A. Cloud Monitoring Uptime Checks with multiple regions selected.
Cloud Monitoring Uptime Checks probe endpoints from multiple global locations and can trigger alerts on failures. Cron jobs don't provide multi-region coverage or integrated alerting.
Question 3 General

You need to provision 50 identical web servers across three zones in us-central1 with automatic scaling based on CPU utilization and automatic replacement of failed instances. What resource should you create?

  • ✓ Managed Instance Group (MIG) with autoscaling and health checks.
  • ○ 50 individual Compute Engine instances with startup scripts.
  • ○ Unmanaged Instance Group with manual scaling.
  • ○ GKE cluster with 50 pods.
Answer: A. Managed Instance Group (MIG) with autoscaling and health checks.
Managed Instance Groups provide automatic scaling, auto-healing (replacing failed instances), rolling updates, and load balancing integration. They're designed for managing fleets of identical VMs.
Question 4 General

Mountkirk Games receives a court order to hold all logs related to a specific player's IP address for potential litigation. The logs are in Cloud Logging. You need to ensure these logs are not deleted by the standard retention policy. What feature should you use?

  • ✓ Create a Log Sink to a locked Bucket.
  • ○ Apply a Tag to the logs.
  • ○ There is no way to hold logs in Cloud Logging.
  • ○ Export to BigQuery.
Answer: A. Create a Log Sink to a locked Bucket.
Standard Cloud Logging buckets have a set retention. To 'Hold' logs indefinitely or for a legal hold, you Export (Sink) them to Cloud Storage and apply a Bucket Lock (Retention Policy) or just keep them there. You cannot 'freeze' specific log entries inside the ephemeral Logging buffer itself easily without moving them.
Question 5 General

EHR Healthcare's CI/CD pipeline deploys to Cloud Run. You want to ensure that 10% of traffic goes to the new revision 'v2' while 90% stays on 'v1' to monitor for errors before a full rollout. How do you achieve this?

  • ○ Use Cloud Deploy to manage the rollout with a canary strategy.
  • ○ Manually edit the Traffic settings in Cloud Run console to split 90/10.
  • ○ Use Istio VirtualService.
  • ✓ Both A and B are valid approaches.
Answer: D. Both A and B are valid approaches.
Cloud Run natively supports traffic splitting between revisions, allowing for granular control during rollouts. While Cloud Deploy manages the delivery pipeline, the splitting itself is an inherent feature of the Cloud Run service.
Question 6 General

Your organization wants to prevent any project from creating a default VPC network. This must be enforced across the entire organization to prevent accidental non-compliant networks. What should you do?

  • ✓ Use an Organization Policy with the constraint `compute.skipDefaultNetworkCreation` set to Enforce.
  • ○ Use a Hierarchical Firewall Policy to block all traffic on default networks.
  • ○ Use VPC Service Controls to block access to the Compute API.
  • ○ Use a Custom IAM Role that denies the `compute.networks.create` permission.
Answer: A. Use an Organization Policy with the constraint `compute.skipDefaultNetworkCreation` set to Enforce.
The Organization Policy Service is designed to enforce constraints across the resource hierarchy. The specific constraint `compute.skipDefaultNetworkCreation` prevents the automatic creation of the 'default' VPC network when new projects are created.

Open the full GCP PCA question-and-answer guide