Examkin

Question

Difficulty: Very hardDesigning Infrastructure for Business Requirements and Cost Optimization

A global healthcare organization is designing a cloud architecture to process batch genomic dataset archives uploaded by partner laboratories. The dataset uploads occur unpredictably throughout the day, with volume fluctuating between zero and several hundred uploads per hour. Each processing task runs a containerized data-transformation binary that takes between 10 and 40 minutes to complete. The business goal is to minimize operational overhead and compute costs by avoiding charges for idle infrastructure while ensuring processing starts within minutes of upload. Additionally, raw uploaded archives must be preserved for compliance for five years, but are rarely accessed after the initial 30 days. Which TWO architectural recommendations should you make to meet these business and cost requirements? (Select TWO.)

  1. Execute containerized processing workloads using Cloud Run jobs triggered automatically by Cloud Storage object creation events via Eventarc.Answer
  2. Configure an Object Lifecycle Management rule on the Cloud Storage bucket to transition objects from Standard to Coldline storage 30 days after creation.Answer
  3. C
    Provision a continuous Google Kubernetes Engine (GKE) Autopilot cluster to run Kubernetes Job objects whenever new archives arrive.
  4. D
    Purchase 3-year Compute Engine Committed Use Discounts (CUDs) to cover the maximum anticipated vCPU and memory capacity required during peak processing hours.
  5. E
    Deploy a multi-region Cloud Spanner instance to store temporary task execution metadata and status logs during archive processing.

Answer

The optimal architecture combines Cloud Run jobs triggered by Eventarc upon Cloud Storage uploads with Cloud Storage Object Lifecycle Management transitioning archives to Coldline storage after 30 days.
Executing batch tasks using Cloud Run jobs triggered by Cloud Storage events allows the compute layer to scale to zero when no uploads occur, meeting the business objective of paying only for active processing time. Simultaneously, applying Cloud Storage Object Lifecycle Management to transition raw archives to Coldline storage after 30 days minimizes five-year retention costs for infrequently accessed data.

Step-by-Step Solution

1
Analyze compute cost requirements for unpredictable batch processing
Identified that workloads fluctuate from zero to hundreds of tasks per hour and require execution times up to 40 minutes.
Cloud Run jobs support tasks running up to 24 hours while scaling completely to zero when idle, avoiding GKE cluster overhead and fixed compute provisioning costs.
2
Analyze storage lifecycle and cost optimization requirements
Identified that uploaded raw files are rarely accessed after 30 days but must be retained for 5 years.
Transitioning objects from Standard to Coldline storage after 30 days significantly reduces long-term storage expenditure while maintaining compliance readiness.
3
Evaluate distractors against FinOps and architecture guidelines
Rejected standard CUD commitments, GKE clusters, and Cloud Spanner metadata storage.
These alternatives introduce unnecessary baseline provisioned costs and over-engineered compute/database capacity for episodic workloads.

Key Concept

Serverless containerized batch architecture with automated storage tiering for unpredictable workloads
Rate this question

Topics

Designing Infrastructure for Business Requirements and Cost Optimization