Prepare and test your skills

Prepare and test your skills

Free sample question with answers and detailed explanation below. The Professional Cloud Developer practice bank has an additional 608 practice questions.Try 10 free realistic questions: instant score, no credit card, and no account required.
Your team is deploying a latency-sensitive financial pricing service that loads large in-memory reference datasets at startup, requiring roughly 45 seconds to initialize.
The workload has the following requirements:
Which platform and scaling strategy should you choose?
Cloud Run with request-based billing and min-instances set to 0 is a fully managed serverless container platform designed for event-driven and intermittent workloads. It allocates CPU only while processing incoming HTTP requests and automatically scales container instances down to zero when traffic stops.
The application were a lightweight, fast-starting microservice (initializing in milliseconds rather than 45 seconds) with sporadic or unpredictable traffic that could tolerate occasional cold starts in exchange for zero idle costs overnight.
Serverless request-based scaling excels at cost-efficiency for bursty, fast-booting HTTP APIs, whereas workloads with heavy memory initialization and strict latency SLAs benefit more from VM-based standby pools and predictive autoscaling.
Configuring Cloud Run with min-instances above zero and instance-based billing keeps a fixed baseline of container instances warm and allocates CPU continuously, even between incoming requests. Adaptive Concurrency Tuning (ACT) automatically adjusts each instance's maximum concurrent requests based on observed CPU usage.
min-instances only keeps the baseline containers warm. When sudden traffic spikes exceed that baseline, Cloud Run must scale out on demand, exposing overflow requests to the 45-second initialization or pending-queue latency.min-instances prevents cold starts for steady baseline traffic, but incurs continuous billing for those warm containers even when idle overnight.The service started up rapidly (under a second), required zero infrastructure management, needed continuous CPU for lightweight background threads, and had predictable peak traffic covered mostly by min-instances without requiring custom VM hardware ratios.
Many architects assume min-instances eliminates all serverless cold starts; in reality, it only protects baseline traffic, whereas MIG standby pools preserve initialized state for rapid spike capacity without full active VM billing.
Google Kubernetes Engine (GKE) Standard provides full control over Kubernetes node pools and Pod scheduling. Pairing Cluster Autoscaler with Spot VMs dynamically adds and removes discounted, preemptible Compute Engine nodes based on pending Pod resource demands.
The workload were a fault-tolerant batch processing job or stateless background pipeline that could tolerate node interruptions, checkpoint progress, and accept several minutes of node provisioning delay in exchange for maximum compute cost savings.
Using Spot VMs for latency-sensitive serving workloads is a common cost-optimization anti-pattern; predictable serving tiers require standard or reserved Compute Engine capacity with standby pools or dedicated node pools.
A Compute Engine Managed Instance Group (MIG) manages a fleet of identical virtual machines (VMs) as a single entity, built from custom OS images and instance templates. When combined with custom machine types, predictive autoscaling, and standby pools, it provides persistent VM-based compute that anticipates recurring demand while keeping pre-initialized instances suspended for rapid scale-out.
Unlike Cloud Run, which faces cold-start or pending-queue delays when scaling beyond warm baseline instances for slow-initializing apps, or GKE with Spot VMs, which risks sudden node preemption, a Compute Engine MIG with predictive autoscaling and standby pools delivers VM-level predictability, zero initialization penalty on spike scale-out, and strong off-peak cost control.
Answer them yourself and get your score the moment you finish. No account needed.
Full access includes
Keep the momentum going with these hand-picked practice scenarios
Last updated Validated