Examkin

Soru

Zorluk: OrtaPlanning and Assessing Cloud Compute Solutions

A smart grid technology company is planning compute resources on Google Cloud for two distinct backend workload components:

1. Workload 1: A stateless REST API endpoint that receives unpredictable, bursty HTTP requests from smart meters. It requires automatic scaling down to zero instances during zero-traffic periods to eliminate idle resource costs.
2. Workload 2: A daily 4-hour batch data transformation pipeline that is completely stateless, fault-tolerant, and capable of saving state checkpoints externally to handle instance interruptions.

Which TWO compute service configurations should the cloud engineer select to satisfy these requirements cost-effectively?

  1. Deploy Workload 1 on Cloud Run to leverage containerized automatic scaling to zero instances when no HTTP traffic is present.Cevap
  2. Provision Compute Engine Spot VMs for Workload 2 to significantly reduce compute costs for short-lived, fault-tolerant batch processing.Cevap
  3. C
    Deploy Workload 1 on Cloud Functions by configuring multi-container Docker Compose definitions for custom runtime dependencies.
  4. D
    Provision standard non-preemptible Compute Engine VMs with 3-year Committed Use Discounts for Workload 2 to prevent compute node preemption during execution.
  5. E
    Deploy Workload 2 on GKE Autopilot clusters configured with custom OS kernel modifications to enable high-throughput batch task processing.

Cevap

The correct architecture decisions are deploying Workload 1 on Cloud Run to enable scale-to-zero capabilities for bursty HTTP requests, and provisioning Spot VMs on Compute Engine for Workload 2 to cost-effectively run fault-tolerant batch workloads.
Cloud Run is the optimal compute service for Workload 1 because it automatically scales containerized HTTP workloads down to zero when no traffic arrives, eliminating idle costs. For Workload 2, Spot VMs on Compute Engine provide massive cost reductions for batch jobs that are stateless and checkpointed to handle potential instance preemptions.

Adım Adım Çözüm

1
Analyze Workload 1 requirements
Workload 1 is a stateless HTTP REST API with bursty, unpredictable traffic that requires scale-to-zero when inactive.
Cloud Run natively supports containerized stateless microservices with automatic request-based scaling down to 0 instances, making it the most cost-efficient choice for bursty web APIs.
2
Analyze Workload 2 requirements
Workload 2 runs 4 hours daily, is stateless, fault-tolerant, and supports checkpointing.
Spot VMs provide significant cost savings (60–91%) for workloads that can tolerate preemption and run for short durations.
3
Evaluate invalid alternative options
Cloud Functions does not support multi-container stacks, Committed Use Discounts are inefficient for 4-hour daily tasks, and GKE Autopilot locks node OS configurations.
Selecting appropriate GCP compute options requires matching workload statefulness, scaling characteristics, and administrative boundaries against GCP product limitations.

Anahtar Kavram

Selecting GCP Compute Services Based on Workload Characteristics and Cost Efficiency
Bu soruyu puanla

Konular

Planning and Assessing Cloud Compute Solutions

Bu konudan daha fazla soru