Examkin

Soru

Zorluk: Çok zorEstimating GCP Costs with the Pricing Calculator

An enterprise architecture team is designing a cost estimation model in the Google Cloud Pricing Calculator for a nightly batch analytics pipeline. The workload requires 50 Compute Engine instances (`n2-standard-16`) operating continuously for 12 hours each night, totaling 365 hours per month per instance. The processing engine is stateless, fault-tolerant, and designed to resume from checkpoints without data loss if an instance is terminated unexpectedly. During the 12-hour operating window, the pipeline processes and updates 20 TB of active data daily, which is continuously read and written to Cloud Storage. Which parameter combination in the Pricing Calculator produces the most accurate configuration for achieving the lowest valid monthly cost estimate?

  1. Configure Compute Engine instances as Spot VMs running for 365 hours per month, and set the Cloud Storage bucket class to Standard storage.Cevap
  2. B
    Configure standard Compute Engine instances for 730 hours per month with a 3-Year Committed Use Discount (CUD), and set the Cloud Storage bucket class to Coldline storage.
  3. C
    Configure Compute Engine instances as Spot VMs with an added 1-Year Committed Use Discount (CUD) applied in the calculator settings, and set the Cloud Storage bucket class to Archive storage.
  4. D
    Configure standard Compute Engine instances for 365 hours per month relying on Sustained Use Discounts (SUDs) for maximum compute savings, and set the Cloud Storage bucket class to Nearline storage.

Cevap

Configure Compute Engine instances as Spot VMs running for 365 hours per month, and set the Cloud Storage bucket class to Standard storage.
Selecting Spot VMs configured for 365 hours per month accurately reflects the part-time, fault-tolerant nature of the nightly batch job while capturing 60-90% compute cost reductions. Paired with Standard Cloud Storage, this avoids data retrieval fees for the 20 TB of data accessed and modified daily.

Adım Adım Çözüm

1
Evaluate Compute Engine instance provisioning type and duration based on workload resilience.
Because the batch workload is stateless, checkpoint-enabled, and fault-tolerant, Spot VMs (which offer 60-90% discounts) are the optimal choice. The hours parameter must be set to 365 hours/month (12 hours/day * 30.4 days).
Spot VMs significantly reduce compute costs compared to standard VMs or full-month CUDs when workloads can tolerate interruptions.
2
Evaluate discount combination rules in the GCP Pricing Calculator.
Committed Use Discounts (CUDs) require a continuous 1-year or 3-year commitment and cannot be stacked on top of Spot VM pricing.
Spot pricing already provides deep discounts; stacking CUDs on Spot instances is invalid in GCP pricing logic.
3
Evaluate Cloud Storage class based on access patterns and retrieval fees.
Standard Cloud Storage is required because 20 TB of data is actively read and written every single day.
Coldline, Nearline, and Archive classes charge per-GB data retrieval fees and have minimum retention periods. Daily active usage in colder tiers would incur high retrieval penalties exceeding any storage space savings.

Anahtar Kavram

Selecting optimal Compute Engine pricing models (Spot vs CUD vs SUD) and Cloud Storage classes based on runtime and data access patterns in the GCP Pricing Calculator.
Tahmini Süre:3m 0s
Bu soruyu puanla

Konular

Estimating GCP Costs with the Pricing Calculator

Bu konudan daha fazla soru