GCP GCP-PCA Free Practice Questions — Page 2

Correct Answers: A. Introduce a green-blue deployment model; C. Fragment the monolithic platform into microservices

A. Introduce a green-blue deployment model A Blue-Green deployment is a strategy that utilizes two identical production environments. How it works: One environment (Blue) handles all live production traffic. You deploy the new version of your application to the second environment (Green). Once you've verified that the Green environment is functioning correctly, you switch the router or load balancer to point all traffic to Green. Why it reduces rollbacks: If a critical error is discovered immediately after the switch, "rolling back" is as simple as flipping the switch back to the Blue environment. There is zero downtime and no need to uninstall software or restore databases in a panic. C. Fragment the monolithic platform into microservices Moving from a Monolith to Microservices addresses the structural risk of deployments. How it works: You break the large, single application into smaller, independent services that communicate via APIs. Each service can be developed, tested, and deployed independently. Why it reduces rollbacks: In a monolith, a single bug in a minor feature can force a rollback of the entire platform. With microservices, an error is isolated to a specific service. You only need to roll back that specific service, while the rest of the platform remains stable and functional. This significantly reduces the blast radius of any single failed deployment.

Correct Answers: A. Use the - -no-auto-delete flag on all persistent disks and stop the VM; D. Use Google BigQuery billing export and labels to associate cost to groups

Using --no-auto-delete on persistent disks (Option A) ensures the disk persists when the VM is stopped, preserving state without paying for a running VM. Stopping (not terminating) the VM avoids compute charges while keeping the disk intact. Option D — BigQuery billing export with labels — gives the finance team cost visibility per team or project. Option B (--auto-delete) would destroy the disk on termination, losing state. Options E and F introduce unnecessary complexity for a dev environment. Option C (CPU utilization label) doesn't provide billing visibility.

Correct Answer: C. Relational

A relational database is the correct choice here. The data involves multiple related entities: sensors, rooms, account owners, and office locations — all of which benefit from JOIN operations and structured queries. The data is structured with discrete fields, not unstructured blobs. Cloud SQL or Cloud Spanner supports this model well. NoSQL (Option B) is better for unstructured or high-velocity data without complex relationships. Flat files (Option A) have no query capability. Blobstore (Option D) is for binary/unstructured object storage, not transactional data.

Correct Answer: C. Ensure that a firewall rule exists to allow load balancer health checks to reach the instances in the instance group.

When instances have no public IP and are being terminated/restarted repeatedly, the most likely cause is failing health checks. The HTTP(S) load balancer sends health check probes from specific Google IP ranges (130.211.0.0/22 and 35.191.0.0/16). If a firewall rule doesn't allow these probes to reach the instances, health checks fail, and the load balancer marks instances as unhealthy and removes them. Option A is about client-to-LB traffic, not the issue here. Option B is unnecessary and contradicts the private-IP design. Option D is not how firewall rules for load balancers work.

Correct Answer: C. Create a new service account with BigQuery access and execute your script with that user

Why Option C is the Best Practice In Google Cloud, applications should use Service Accounts to authenticate. This follows the Principle of Least Privilege. Identity: You create a specific identity (the Service Account) for the script. Permissions: You grant that account only the specific IAM roles it needs (e.g., roles/bigquery.dataEditor or roles/bigquery.jobUser). Portability: This method works regardless of where the script runs (GCE, local machine, or another cloud) and doesn't require recreating the VM.