Examkin

Question

Difficulty: MediumOptimizing Business Processes through FinOps and Cloud Cost Management

An enterprise online food delivery platform is optimizing its Google Cloud footprint as part of a new FinOps governance initiative. The platform operates a steady-state core order-dispatch service on Compute Engine VMs that runs continuously 24/7 with minimal traffic variance. In addition, the platform runs highly unpredictable, bursty machine learning batch jobs for real-time demand forecasting that execute intermittently for 2 to 4 hours a day. The platform engineering team needs to minimize infrastructure costs while establishing clear cost attribution per microservice team. Which cloud architecture and billing strategy should the Cloud Architect recommend?

  1. A
    Purchase 3-year standard Committed Use Discounts (CUDs) sized to cover the peak capacity of both the steady-state Compute Engine instances and the bursty machine learning batch jobs.
  2. B
    Migrate both the steady-state dispatch service and the intermittent batch workloads onto a dedicated Google Kubernetes Engine (GKE) Enterprise cluster configured with static node pools to simplify billing management.
  3. Purchase 3-year Flexible Committed Use Discounts (CUDs) for the steady-state Compute Engine instances, run the unpredictable batch ML workloads on Spot VMs, and enforce resource labeling linked to Cloud Billing export for cost attribution.Answer
  4. D
    Grant the primitive Owner role to the FinOps lead across all GCP projects so they can manually delete unutilized compute instances and adjust project billing settings.

Answer

Purchase 3-year Flexible Committed Use Discounts (CUDs) for baseline compute, leverage Spot VMs for fault-tolerant bursty batch processing, and apply granular resource labels tied to BigQuery Cloud Billing exports for cost allocation.
Combining Flexible Committed Use Discounts (CUDs) for predictable 24/7 baseline infrastructure with Spot VMs for fault-tolerant, intermittent batch jobs ensures maximum cost efficiency. Additionally, enforcing resource labels linked to Cloud Billing exports satisfies FinOps cost attribution requirements across different application teams.

Step-by-Step Solution

1
Analyze workload characteristics for compute cost optimization.
Identified steady-state 24/7 compute requirements alongside intermittent, bursty batch workloads.
Steady-state workloads benefit from term commitments (CUDs), whereas bursty, fault-tolerant workloads achieve maximum savings through preemptible/Spot VMs.
2
Select appropriate Google Cloud discount models.
Apply Flexible CUDs to baseline Compute Engine instances and utilize Spot VMs for the demand forecasting batch jobs.
Flexible CUDs offer guaranteed discounts on hourly spend for stable compute, while Spot VMs reduce compute costs by up to 60-91% for batch processing.
3
Implement FinOps governance and cost attribution mechanisms.
Enforce key-value resource labels across infrastructure resources and export detailed billing logs to BigQuery.
Labels provide micro-level visibility into spend by team, environment, or service, enabling accurate showback/chargeback reporting.

Key Concept

FinOps Cost Optimization and Governance
Rate this question

Topics

Optimizing Business Processes through FinOps and Cloud Cost Management