Examkin

Question

Difficulty: HardPlanning Compute Engine Resources and Machine Types

An enterprise company is planning to migrate two distinct workloads to Google Compute Engine. Workload 1 is a mission-critical relational database that runs continuously 24/7, requires 128 vCPUs and 512 GB RAM, and cannot tolerate unexpected terminations. Workload 2 is a batch data processing job that runs overnight, is designed to resume from checkpoints if interrupted, and needs to be executed as cost-effectively as possible. Which resource allocation and provisioning strategy aligns with Google Cloud recommended practices?

  1. Provision standard compute instances with Committed Use Discounts (CUDs) for Workload 1, and use Spot VMs for Workload 2.Answer
  2. B
    Provision Spot VMs for both Workload 1 and Workload 2 to minimize compute expenses across all environments.
  3. C
    Purchase upfront Sustained Use Discounts for Workload 2, and deploy Workload 1 on Preemptible VMs attached to 3-year Committed Use Discounts.
  4. D
    Deploy Workload 1 to Cloud Functions for automated horizontal scaling, and allocate dedicated sole-tenant nodes for Workload 2.

Answer

Provision standard compute instances with Committed Use Discounts for the continuous stateful workload, and use Spot VMs for the fault-tolerant batch processing workload.
The option recommending Committed Use Discounts for the 24/7 mission-critical database and Spot VMs for the checkpoint-enabled batch processing job correctly pairs workload fault tolerance with Google Cloud pricing models. Committed Use Discounts reduce costs for predictable 24/7 compute usage without preemption risk, while Spot VMs minimize batch costs for workloads that handle interruptions gracefully.

Step-by-Step Solution

1
Analyze the requirements and fault tolerance of Workload 1.
Workload 1 runs 24/7, requires dedicated memory/CPU scale, and cannot tolerate interruptions.
Stateful database workloads require predictable availability, making Spot or Preemptible instances unsuitable.
2
Evaluate discounting models for steady-state workloads.
Committed Use Discounts (CUDs) provide up to 57% (or 70% for memory-optimized) savings for predictable continuous usage over a 1-year or 3-year term.
Because Workload 1 runs continuously, committing to vCPU and memory resource usage maximizes cost savings without compromising instance stability.
3
Analyze the requirements and fault tolerance of Workload 2.
Workload 2 is a batch process that can resume from checkpoints if interrupted.
Stateless or checkpoint-enabled batch jobs can tolerate preemption, making them ideal candidates for Spot VMs, which offer 60-91% discounts compared to standard pricing.

Key Concept

Selecting Compute Engine machine provisioning models and discounting strategies based on workload availability demands and fault tolerance.
Rate this question

Topics

Planning Compute Engine Resources and Machine Types

More questions from this topic