Google Cloud (GCP) Quiz

Google Cloud Platform services and concepts — compute, storage, and the GCP-specific tooling teams rely on.

This category currently has 100 questions in the SERVBG quiz bank. Below are a few sample questions — the full interactive quiz shuffles through the whole set with instant scoring.

Sample questions

A Managed Instance Group (MIG) is configured with autoscaling based on CPU utilization. During a sudden traffic spike, new instances are not being created fast enough. Which MIG feature reduces scale-out latency by keeping pre-initialized instances ready?

Switch the autoscaling metric from CPU utilization to HTTP load balancing serving capacity
Enable stateful MIG configuration to preserve instance state and allow faster instance re-use
Enable live migration on the MIG so existing instances expand their allocated CPU during spikes
Increase the cooldown period so autoscaling decisions are made more aggressively during traffic spikes
A MIG standby policy with a target-size surplus keeps provisioned-but-suspended instances that can be activated without full boot time

You need to run a batch workload on GCE Spot VMs. The job must checkpoint every 30 seconds and tolerate preemption with no more than 30 seconds of lost work at lowest cost. Which combination achieves this?

Use Spot VMs with a shutdown script writing checkpoints to Cloud Storage on SIGTERM, and configure the MIG to restart preempted instances automatically
Use Spot VMs and store checkpoints in instance memory, relying on the 60-second preemption notice
Use Spot VMs with live migration enabled to prevent any data loss during host preemption
Use standard N2 VMs with committed-use discounts and local SSD for checkpoint storage
Use Spot VMs with a startup script that polls for preemption metadata and pauses work proactively

Which statement accurately describes the primary operational difference between GKE Autopilot and GKE Standard clusters?

Autopilot disables all node-level SSH access and Standard clusters enable it by default
Autopilot clusters run on bare-metal nodes while Standard clusters run on virtual machines
Autopilot supports only Deployment workloads while Standard supports StatefulSets and DaemonSets
Autopilot requires manual node pool creation but automates control-plane upgrades only
In Autopilot, Google manages node provisioning, scaling, and OS upgrades; billing is per Pod resource request rather than per node

GKE Enterprise was announced as a rebranding of which previous Google Cloud product portfolio in 2023, and what is its primary addition over standard GKE?

GKE Enterprise rebranded Kubernetes Engine Marketplace in 2023, bundling certified third-party operator subscriptions
GKE Enterprise rebranded Anthos in 2023, adding multi-cluster fleet management, Config Sync, and Policy Controller across hybrid and multi-cloud environments
GKE Enterprise rebranded GKE Autopilot in 2022, adding dedicated bare-metal node support and GPU scheduling primitives
GKE Enterprise rebranded Cloud Run for Anthos in 2021, adding serverless node autoscaling to standard GKE clusters
GKE Enterprise rebranded Borg-as-a-Service in 2023, providing direct access to Google-internal scheduling APIs

Cloud Run is configured with min-instances set to 5 and max-instances set to 20. Traffic drops to zero. What happens to the running instances?

Cloud Run sends SIGTERM to the 5 minimum instances and provisions fresh containers on the next request
All 20 instances are retained until the next deployment to avoid cold starts
The service scales down to the 5 minimum instances, which remain allocated and warm so requests are handled without cold starts
All instances scale to zero regardless of min-instances when no requests arrive for 15 minutes
The service retains a random sample of instances between 1 and 5 based on available capacity in the region

Start the full quiz ›