Examkin

Question

Difficulty: HardSelecting and Designing Compute and Application Platforms

A global media streaming platform is designing the architecture for a new stateless image processing service on Google Cloud. The service processes incoming HTTP requests containing user uploaded images, converts them into multiple web formats, and returns the result. The workload experiences unpredictable traffic spikes ranging from thousands of requests per second during major events to long periods of complete inactivity overnight. To align with FinOps goals and minimize engineering maintenance, the platform team requires a serverless compute model that automatically scales down to zero instances when idle, charges only during request execution, and requires zero cluster administration or node maintenance. Which compute platform should you recommend?

  1. Deploy the containerized service on Cloud Run with request-based autoscaling enabled.Answer
  2. B
    Deploy the service on a Google Kubernetes Engine (GKE) Autopilot cluster using the Horizontal Pod Autoscaler.
  3. C
    Deploy the application across a Compute Engine Managed Instance Group (MIG) using custom VM images and CPU utilization autoscaling.
  4. D
    Deploy the application on Compute Engine Spot VMs with 3-year Committed Use Discounts (CUDs) to cover baseline processing capacity.

Answer

Deploying the containerized service on Cloud Run with request-based autoscaling is the optimal platform choice because Cloud Run provides serverless container execution that automatically scales down to zero instances during idle periods, eliminates cluster management overhead, and bills strictly for active request processing time.
Deploying on Cloud Run fulfills all functional and operational requirements: it natively runs containerized stateless HTTP applications, automatically scales pods up during traffic spikes, scales down to zero during idle overnight hours, requires zero infrastructure or cluster maintenance, and charges strictly per millisecond of request execution.

Step-by-Step Solution

1
Analyze workload characteristics and technical requirements.
Identified stateless HTTP workload, containerized packaging, unpredictable traffic spikes, and requirement for scaling to zero.
Selecting the correct compute option requires evaluating statefulness, invocation triggers, and operational constraints.
2
Evaluate operational overhead and scaling capabilities of GCP compute services.
Cloud Run offers zero infrastructure administration and native scale-to-zero capabilities, matching the requirements exactly.
GKE and Compute Engine MIGs introduce cluster management, node patching, or baseline running costs that violate the zero-administration and zero-idle-cost constraints.
3
Verify cost efficiency alignment against FinOps requirements.
Cloud Run request-based billing ensures no charges accumulate overnight during periods of total inactivity.
Committed Use Discounts or fixed VM pools create unutilized infrastructure costs when traffic drops to zero.

Key Concept

Compute Platform Selection for Stateless Serverless Workloads
Estimated Time:2m 0s
Rate this question

Topics

Selecting and Designing Compute and Application Platforms