Examkin

Question

Difficulty: MediumPlanning Compute Engine Resources and Machine Types

A bioinformatics research organization is planning its compute architecture on Google Cloud for two upcoming workloads:
1. Genomic Pipeline: Interruptible, stateless batch jobs that process large genomic dataset files overnight and can resume from checkpoints if interrupted.
2. Patient Portal API: A continuous 24/7 web application backend with predictable, steady-state resource utilization throughout the year.

Which TWO strategies should the cloud engineer implement to optimize cost and performance for these Compute Engine resources?

  1. Provision Spot VMs for the genomic processing batch jobs to achieve significant cost savings on fault-tolerant workloads.Answer
  2. Purchase Committed Use Discounts (CUDs) for the vCPU and memory baseline required by the patient portal API.Answer
  3. C
    Provision Spot VMs for the relational database backend of the patient portal API to reduce infrastructure costs.
  4. D
    Deploy the patient portal API legacy multi-container workload onto Cloud Functions to minimize compute overhead.

Answer

Provision Spot VMs for the genomic processing batch jobs and purchase Committed Use Discounts (CUDs) for the patient portal API vCPU and memory baseline.
The correct strategy combines Spot VMs for interruptible batch workloads and Committed Use Discounts (CUDs) for continuous steady-state baseline workloads. Spot VMs deliver deep discounts for batch jobs that tolerate preemption, while CUDs provide predictable cost savings for 24/7 web API workloads without preemption risks.

Step-by-Step Solution

1
Analyze the Genomic Pipeline workload requirements.
Identified as a stateless, fault-tolerant batch processing job with checkpointing support.
Fault-tolerant workloads can withstand instance preemption, making them ideal candidates for Spot VMs to maximize cost savings.
2
Analyze the Patient Portal API workload requirements.
Identified as a continuous 24/7 steady-state production service.
Predictable continuous baseline workloads benefit most from Committed Use Discounts (CUDs) without risk of service disruption.
3
Evaluate and reject anti-patterns.
Spot VMs are inappropriate for stateful databases, and Cloud Functions cannot run multi-container legacy applications directly.
Database backends require steady availability, and multi-container workloads exceed Cloud Functions architectural scope.

Key Concept

Planning Compute Engine Workload Cost Optimization Strategy
Rate this question

Topics

Planning Compute Engine Resources and Machine Types

More questions from this topic