Soru

Zorluk: OrtaPlanning Compute Engine Resources and Machine Types

A digital media engineering team needs to run long-running video transcoding processing tasks using a custom compiled C++ binary on Google Cloud. The workload requires maximum single-thread CPU clock speeds, consistent processor performance, and runs uninterrupted in contiguous 12-hour job windows. If a node terminates mid-job, the entire 12-hour transcoding process fails and must restart from the beginning. Which Compute Engine resource planning strategy should the cloud engineer recommend?

  1. Provision Compute-optimized (C2 or C3) machine family instances running as standard Virtual Machines.Cevap
  2. B
    Provision Compute-optimized (C2 or C3) machine family instances running as Spot Virtual Machines to reduce compute costs.
  3. C
    Deploy the custom compiled binary using Cloud Functions triggered by Cloud Storage file uploads.
  4. D
    Provision E2 shared-core machine types and rely on automatic Sustained Use Discounts (SUDs) to minimize costs.

Cevap

Provision Compute-optimized machine family instances (such as C2 or C3) configured as standard Compute Engine virtual machines.
Compute-optimized machine types (C2 and C3) provide high per-core CPU performance and high clock frequencies tailored for CPU-intensive compute workloads like video encoding. Using standard VMs ensures uninterrupted execution, satisfying the strict requirement that the 12-hour process must not be interrupted.

Adım Adım Çözüm

1
Analyze workload compute profile requirements.
The application requires high single-thread CPU performance and sustained high clock frequency for CPU-bound video transcoding tasks.
Matching CPU performance profiles to Google Cloud machine families requires picking Compute-optimized (C2/C3) machines rather than General-purpose or Memory-optimized types.
2
Evaluate fault tolerance and availability constraints.
The job runs continuously for 12 hours and fails completely if interrupted.
Because the workload cannot recover from sudden termination, Spot/Preemptible VMs must be ruled out in favor of standard Compute Engine VMs.
3
Verify execution environment limits for alternative cloud offerings.
Serverless functions (Cloud Functions) cannot be used due to strict 9-minute execution timeouts.
Long-running batch workloads exceeding serverless timeouts must run on Compute Engine virtual instances.

Anahtar Kavram

Compute Engine Machine Family Selection and Spot VM Suitability
Tahmini Süre:1m 30s
Bu soruyu puanla