Reliability & DR

High availability, disaster recovery, and failover strategies

1

Set 1

20 questions • ~25 min

Start Set 1
2

Set 2

12 questions • ~25 min

Start Set 2

All Questions

32 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

You have a set of Cloud Run services. You want to control access so that only specific authorized users can invoke them. You want to use a centralized policy. What do you use?

  • ✓ Identity-Aware Proxy (IAP) with an External Load Balancer.
  • ○ VPC Service Controls.
  • ○ Cloud Armor.
  • ○ Allow unauthenticated, use internal logic.
Answer: A. Identity-Aware Proxy (IAP) with an External Load Balancer.
IAP allows you to enforce access control (User Identity) at the Load Balancer layer. By placing Cloud Run behind a GLB (Serverless NEG) and enabling IAP, you can centrally manage 'who' can access the app (e.g. 'All Employees').
Question 2 General

You're paying for Cloud Interconnect capacity that's underutilized. Traffic uses only 20% of provisioned bandwidth. What optimization reduces costs?

  • ✓ Downsize the Interconnect capacity or consider Partner Interconnect for lower bandwidth needs.
  • ○ Increase Interconnect capacity.
  • ○ Add more Interconnect attachments.
  • ○ Interconnect capacity cannot be changed.
Answer: A. Downsize the Interconnect capacity or consider Partner Interconnect for lower bandwidth needs.
Partner Interconnect offers smaller capacity increments than Dedicated Interconnect. Right-sizing to actual usage reduces fixed infrastructure costs.
Question 3 General

Your Spark job on Dataproc spills to disk frequently, slowing execution. What optimization helps?

  • ✓ Increase executor memory or optimize partitioning to reduce shuffle size.
  • ○ Reduce memory further.
  • ○ Disk spill is optimal.
  • ○ Use smaller datasets only.
Answer: A. Increase executor memory or optimize partitioning to reduce shuffle size.
Disk spill indicates insufficient memory for operations. Increasing executor memory or reducing shuffle size through better partitioning addresses this.
Question 4 General

Mountkirk Games wants to minimize latency for their First Person Shooter. They found that traversing the public internet to reach the GCP region is adding jitter. They want to use Google's backbone. What load balancer / network tier is default and recommended?

  • ✓ Premium Tier (passes traffic to Google edge PoP).
  • ○ Standard Tier.
  • ○ Carrier Peering.
  • ○ Direct Peering.
Answer: A. Premium Tier (passes traffic to Google edge PoP).
Premium Tier is the default. It on-ramps traffic at the nearest PoP to the user and carries it on Google's private low-jitter backbone. This is ideal for gaming.
Question 5 General

A company processes sensitive financial transactions and wants to ensure developers can test code in dev/test environments without accessing production data. What data management strategy should you implement?

  • ○ Copy production database to dev/test weekly.
  • ✓ Use Sensitive Data Protection (DLP) to de-identify production data before copying to dev/test.
  • ○ Use the same database with different schemas.
  • ○ Generate synthetic test data only.
Answer: B. Use Sensitive Data Protection (DLP) to de-identify production data before copying to dev/test.
Sensitive Data Protection (Cloud DLP) can automatically detect and de-identify sensitive data while maintaining data utility for testing.
Question 6 General Choose 2

You have a legacy monolithic application that cannot be containerized yet. You want to maximize reliability on Compute Engine. Which TWO features should you use?

  • ✓ Managed Instance Group (MIG) with Autohealing.
  • ✓ Regional Persistent Disks.
  • ○ Preemptible instances.
  • ○ Sole-tenant nodes.
  • ○ Global Load Balancer.
Answer: A. Managed Instance Group (MIG) with Autohealing. | B. Regional Persistent Disks.
For a VM-based monolith, Autohealing (A) ensures the VM is restarted if the app freezes (health check fails). Regional PD (B) ensures that if a Zone fails, the data is safe in another zone and can be force-attached to a new instance in that zone (HA storage).

Open the full GCP PCA question-and-answer guide