PCA Analyze and optimize technical and business processes Practice Questions

Explanation: When a Cloud Function fails to process a Pub/Sub message and retry on failure is set to false, the message is not redelivered. Pub/Sub relies on the subscriber (the Cloud Function) to acknowledge messages; without retries, a failed execution causes the message to be dropped after the ack deadline expires, leading to message loss.

Explanation: The Coldline storage class is designed for data accessed less than once a quarter, making it the most cost-effective choice for data accessed roughly once a quarter. It offers lower storage costs than Standard or Nearline, with retrieval fees and a 90-day minimum storage duration that suits quarterly access patterns.

Explanation: Option A is correct because replicating snapshots to a secondary region ensures that a recent, consistent copy of the regional persistent disk data exists outside the primary region, enabling fast recovery with a low RPO and RTO if the primary region fails. Option E is correct because creating a custom image of the boot disk and copying it to another region allows rapid instance recreation in the secondary region, reducing RTO by avoiding OS and application reinstallation. Option B is wrong because manual snapshots taken only after a failure cannot meet low RPO, since data written since the last snapshot would be lost. Option C is wrong because an instance template in the same region does not provide cross-region resilience and would be unavailable if the region fails. Option D is wrong because Cloud Storage backups with a lifecycle policy are typically for archival or long-term retention and do not by themselves deliver the low RTO/RPO required for a critical regional persistent disk workload.

Explanation: Option A is correct because Bigtable sorts data lexicographically by row key, so shorter row keys reduce the amount of data read and compared per row, lowering scan latency across a large range. Option B is correct because adding nodes increases the cluster's read throughput and CPU capacity, which directly reduces latency for large scans that are throughput-bound. Option E is correct because when a table is replicated across clusters, application profiles with single-cluster routing prevent the client from reading across multiple clusters, avoiding cross-cluster latency and keeping reads local to one fast cluster. Option C is not a performance-improving action by itself; Key Visualizer is a diagnostic tool that identifies hot spots and access patterns but does not directly reduce query latency. Option D is incorrect because HDD storage has higher latency than SSD, so switching from SSD to HDD would worsen query performance rather than improve it.

Explanation: The `gcloud compute instances describe` output shows the instance has an external IP that is ephemeral (not reserved). To create a new instance with the same configuration but a different external IP, the simplest approach is to create a new instance without specifying a static IP, which will automatically assign a new ephemeral IP. This avoids the overhead of reserving and managing a static IP when only a temporary, different address is needed.