PCD Designing highly scalable, available, and reliable cloud-native applications Practice Test 3 — 15 Questions

Question 1

A developer is designing a data pipeline using Pub/Sub and Dataflow. They need to guarantee at-least-once delivery with no duplicates in the sink. Which Dataflow feature should they use?

Accepted Answer

Exactly-once processing. Dataflow's exactly-once processing (also known as 'exactly-once semantics' or 'EOS') ensures that each record is processed exactly once, even if the pipeline restarts or fails. This eliminates duplicates in the sink while still guaranteeing at-least-once delivery from Pub/Sub, because Dataflow uses a combination of source-side deduplication (via Pub/Sub message IDs) and sink-side idempotent writes (via the Dataflow sink's commit protocol). The result is that no duplicate records are written to the sink, meeting the requirement of no duplicates.

Answer

Checkpointing

Answer

Idempotent writes

Answer

Windowing

Question 2

Which three factors should be considered when choosing a regional vs. multi-regional deployment for a globally distributed application?

Accepted Answer

Data residency requirements. Option A (Data residency requirements) is correct because regulations such as GDPR or data-sovereignty laws may mandate that user data stays within specific geographic boundaries, which directly determines whether a multi-regional deployment is even permissible. Option B (Cost of data transfer) is correct because multi-regional architectures incur cross-region and egress data-transfer charges that can significantly raise operating costs compared with a single regional deployment. Option E (Latency for users) is correct because serving users from multiple regions reduces round-trip network latency, which is a primary driver for choosing multi-regional over regional deployment. Option C (Single region compliance) is not a distinct factor — compliance is already captured by data residency, and 'single region compliance' is not a recognized deployment consideration. Option D (Replication lag) is a consequence of multi-regional design rather than a primary selection factor, and it is typically managed through consistency models rather than being a deciding criterion between regional and multi-regional deployment.

Answer

Single region compliance

Answer

Replication lag

Question 3

Which two strategies should be implemented to ensure high availability for a Compute Engine instance group running a stateless web application?

Accepted Answer

Use regional managed instance group. Option B is correct because a regional managed instance group (MIG) automatically spreads its instances across multiple zones within a region and uses the managed instance group's autohealing and rebalancing features to maintain the desired capacity even if a zone fails. Option E is correct because multi-zone deployment ensures the stateless web application runs in more than one zone, so a single zone outage does not take down the entire service; this is the core availability pattern for Compute Engine workloads. Option A is not appropriate because preemptible VMs can be terminated at any time and are not designed for high availability. Option C is not the right fit because global load balancing distributes traffic across regions but does not by itself provide the multi-zone instance redundancy required for high availability. Option D is not correct because instance templates only define the VM configuration used to create instances; they do not provide availability by themselves.

Answer

Use preemptible VMs

Answer

Use global load balancing

Answer

Use instance templates

Question 4

Which two design patterns help decouple microservices?

Accepted Answer

Event-driven architecture. Event-driven architecture (B) decouples microservices because producers publish events to a broker (e.g., Kafka, RabbitMQ) without knowing which consumers exist, so services communicate asynchronously through events rather than direct synchronous calls, reducing runtime dependencies. An API gateway (D) decouples clients from the internal microservice topology by providing a single entry point that handles routing, protocol translation, and cross-cutting concerns, so clients don't need to know individual service locations or contracts. Service mesh (A) is about handling service-to-service communication concerns like mTLS, retries, and observability via sidecar proxies, but it still relies on direct service-to-service calls, so it doesn't fundamentally decouple services. Database per service (C) is a data-management pattern that prevents shared-database coupling, but it addresses data ownership rather than the communication decoupling asked about here. Circuit breaker (E) is a resilience pattern that prevents cascading failures by failing fast, but it presupposes direct calls between services and therefore does not decouple them.

Answer

Service mesh

Answer

Database per service

Answer

Circuit breaker

Question 5

Refer to the exhibit. A developer notices that instance-3 is in TERMINATED state. What is the most likely reason?

Accepted Answer

The instance had automatic restart disabled. When an instance's 'automatic restart' is disabled, the instance will not be automatically restarted after a host maintenance event or a failure. If the underlying host experiences an issue, the instance transitions to TERMINATED state instead of being migrated or restarted. This is the most likely reason for instance-3 being in TERMINATED state while other instances remain running.

Answer

The instance was deleted

Answer

The instance was preempted

Answer

The instance's zone was unavailable

Question 6

Refer to the exhibit. The user developer@example.com tries to create a firewall rule and receives a permission denied error. What is the most likely reason?

Accepted Answer

All of the above. All of the above. In Google Cloud, creating a firewall rule requires the `compute.securityAdmin` role (which includes `compute.firewalls.create` permission) or the `compute.networkAdmin` role (which also includes `compute.firewalls.create` permission). If the user lacks any of these roles or the specific permission, they will receive a permission denied error. While options A, B, and C are individually true statements, the question asks for the 'most likely reason,' and since all three are valid, the single correct answer is D. This tests the understanding that multiple permission deficiencies can cause the same error.

Answer

The user lacks compute.networkAdmin role

Answer

The user lacks compute.securityAdmin role

Answer

The user is missing compute.firewalls.create permission

Question 7

A financial trading application on Compute Engine requires an RPO of 5 seconds and RTO of 1 minute for zone failures. Which architecture should they use?

Accepted Answer

Regional persistent disk attached to a single instance. Regional persistent disks provide synchronous replication of data between two zones within a region, ensuring an RPO of effectively zero (typically under 5 seconds) and enabling rapid failover to a secondary zone. By attaching the regional disk to a single Compute Engine instance, the application can quickly resume operations in the other zone upon failure, meeting the 1-minute RTO without data loss or complex replication overhead.

Answer

Persistent disk with periodic snapshots to a different zone

Answer

Managed instance group with autoscaling and health checks

Answer

Two instances in different zones with data replicated via rsync

Question 8

A stateful service on GKE needs to persist data that must be accessible from any pod in the cluster, regardless of which node the pod runs on. Which volume type should they use?

Accepted Answer

PersistentVolumeClaim with RWX access mode. A PersistentVolumeClaim (PVC) with RWX (ReadWriteMany) access mode is correct because it allows multiple pods across different nodes to read and write to the same persistent volume simultaneously. This is essential for a stateful service where data must be accessible from any pod in the cluster, regardless of which node the pod runs on. RWX is typically backed by network filesystems like NFS or GKE Filestore, which provide shared access across nodes.

Answer

emptyDir

Answer

ConfigMap

Answer

hostPath

Question 9

An application on Cloud Run needs to connect to a Cloud SQL instance securely with minimal latency. It also needs to access Cloud Storage buckets in the same region. Which networking configuration should they use?

Accepted Answer

Serverless VPC Access connector with Private Services Access for Cloud SQL. Serverless VPC Access connector enables Cloud Run to send requests to resources in a VPC network via internal IPs, while Private Services Access allocates an internal IP range for Cloud SQL, allowing direct, low-latency connectivity without traversing the public internet. This combination ensures secure, minimal-latency access to Cloud SQL and also allows Cloud Run to reach Cloud Storage buckets in the same region through the VPC connector's internal routing.

Answer

Cloud NAT for outbound traffic

Answer

Direct VPC peering with Cloud SQL

Answer

Use Cloud SQL Auth Proxy with public IP

Question 10

A multi-region application uses Cloud Spanner. The team needs to ensure that a write is immediately visible to all subsequent reads, even those performed in different regions. Which consistency mode should they use?

Accepted Answer

Strong consistency. Strong consistency (D) ensures that once a write is acknowledged, any subsequent read, regardless of region, will reflect that write. Cloud Spanner uses the TrueTime API and Paxos-based replication to provide external consistency (a form of strong consistency) across regions, making it the correct choice for immediate global visibility.

Answer

Eventual consistency

Answer

Global consistency

Answer

Bounded staleness

Question 11

A company deploys a microservice on Cloud Run and wants to minimize cold starts during traffic spikes. Which two steps should they take? (Select exactly 2.)

Accepted Answer

Enable CPU always allocated. Option A (Enable CPU always allocated) is correct because, by default, Cloud Run throttles CPU to near zero when a container instance is idle; keeping CPU always allocated lets background work and initialization continue so instances stay warm and respond faster when a request arrives. Option E (Set min_instances to a value greater than 0) is correct because it keeps a specified number of container instances running at all times, so incoming requests during a traffic spike are served by already-started instances instead of triggering new cold starts. Option B (Use Cloud CDN) is not correct because CDN caching reduces latency and origin load for cacheable content but does not prevent Cloud Run container cold starts. Option C (Set max_instances to a high value) is not correct because max_instances only caps how many instances can scale out; it does not keep instances warm or reduce cold starts. Option D (Set concurrency to 1) is not correct because limiting each instance to one concurrent request actually forces more instances to be created and can worsen cold-start frequency rather than minimize it.

Answer

Use Cloud CDN

Answer

Set max_instances to a high value

Answer

Set concurrency to 1

Question 12

A company runs a stateful application on Compute Engine. They need to achieve an RPO of less than 15 minutes and an RTO of less than 30 minutes for a regional disaster. Which three steps should they include in their disaster recovery plan? (Select exactly 3.)

Accepted Answer

Develop custom scripts to replicate application data asynchronously to another region. Option B is correct because asynchronously replicating application data to another region ensures a cross-region copy of stateful data exists, which is essential to meet an RPO under 15 minutes and to recover from a regional disaster. Option C is correct because persistent disk snapshots stored in a different region provide durable, cross-region recovery points for Compute Engine disks, directly supporting both the RPO and RTO targets during a regional failure. Option E is correct because Cloud DNS geo-routing (or failover routing) directs users to a healthy region's endpoint, enabling traffic redirection during a regional outage and helping achieve the RTO under 30 minutes. Option A does not belong because a multi-zone managed instance group only protects against zonal failures within a single region, not a regional disaster. Option D does not belong because regional persistent disks replicate only within the same region, so they cannot survive a full regional outage.

Answer

Use a managed instance group in multiple zones within the same region

Answer

Use regional persistent disks to replicate data within the region

Question 13

A startup is deploying a stateless web app on Compute Engine. They expect traffic spikes. What is the most cost-effective way to handle scaling?

Accepted Answer

Use managed instance groups with autoscaling based on CPU utilization.. Managed instance groups (MIGs) with autoscaling based on CPU utilization are the most cost-effective solution for a stateless web app with traffic spikes because they automatically add or remove VM instances in response to real-time CPU load, ensuring you only pay for the compute resources you actually use. This approach directly matches the stateless nature of the app, allowing instances to be created and destroyed without data loss, and avoids over-provisioning or under-utilizing resources.

Answer

Use App Engine Standard.

Answer

Use a single large VM with more cores.

Answer

Use Cloud Functions.

Question 14

A company wants to run a batch job every hour that processes files from Cloud Storage. The job takes about 10 minutes. Which serverless option should they use?

Accepted Answer

Cloud Functions with Cloud Scheduler. Cloud Functions with Cloud Scheduler is a serverless option that can run a batch job every hour. Cloud Scheduler triggers the function on a cron schedule, and the function can process files from Cloud Storage. The job takes about 10 minutes, which is within Cloud Functions' maximum timeout of 60 minutes (for 2nd gen) or 9 minutes (1st gen), but 10 minutes exceeds 1st gen limit, so 2nd gen is suitable. However, the question likely expects Cloud Functions with Cloud Scheduler as the serverless option.

Answer

Cloud Run jobs

Answer

Compute Engine with cron

Answer

App Engine Cron Service with Cloud Tasks

Question 15

A developer is building a Cloud Pub/Sub-based event-driven system. They need to ensure that messages are processed at least once, and they want to handle processing failures. What should they do?

Accepted Answer

Use push subscriptions with a dead-letter topic. Using push subscriptions with a dead-letter topic ensures at-least-once delivery and provides a mechanism to handle processing failures. When a push subscription fails to deliver a message (e.g., due to a downstream error), Pub/Sub automatically retries delivery. After exhausting the maximum delivery attempts (default 5), the message is forwarded to a dead-letter topic, where it can be analyzed or reprocessed without losing the message. This guarantees that every message is either processed successfully or stored for manual intervention, satisfying the at-least-once requirement.

Answer

Use pull subscriptions with auto-acknowledgment

Answer

Configure max delivery attempts on the subscription

Answer

Use Cloud Tasks instead of Pub/Sub