Examkin

Question

Difficulty: MediumPlanning Compute Engine Resources and Machine Types

A retail analytics company is planning its Google Compute Engine infrastructure for two distinct application services:
1. An online, memory-intensive analytics engine requiring large RAM allocations, predictable baseline CPU performance, and continuous 24/7 availability over a multi-year timeframe.
2. A stateless data ingestion pipeline that processes incoming messages asynchronously from a queue and can recover gracefully if individual instances are terminated unexpectedly.

Which TWO deployment strategies should the cloud engineer select to meet operational requirements while optimizing cost efficiency? (Select 2 answers.)

  1. Provision High-Memory or Memory-Optimized VM instances coupled with Committed Use Discounts (CUDs) for the 24/7 analytics engine.Answer
  2. Provision Spot Virtual Machines for the stateless data ingestion pipeline to minimize compute costs.Answer
  3. C
    Provision Spot Virtual Machines for the 24/7 analytics engine to achieve maximum compute discount rates.
  4. D
    Deploy the memory-intensive analytics engine to Cloud Functions to automatically eliminate VM infrastructure management.

Answer

The cloud engineer should select provisioning High-Memory or Memory-Optimized VM instances with Committed Use Discounts (CUDs) for the steady-state analytics engine, and provisioning Spot Virtual Machines for the fault-tolerant data ingestion pipeline.
Provisioning High-Memory or Memory-Optimized instances with Committed Use Discounts fulfills the high-RAM 24/7 availability requirements of the analytics engine while securing deep steady-state discounts. Provisioning Spot VMs for the stateless data ingestion pipeline leverages maximum discounts for a workload that easily handles sudden instance terminations.

Step-by-Step Solution

1
Analyze the operational profile of the analytics engine workload.
Determined to require high RAM allocations and continuous 24/7 multi-year uptime.
High-Memory / Memory-Optimized machine types satisfy the resource ratio, while Committed Use Discounts (CUDs) minimize costs for steady 24/7 utilization.
2
Analyze the operational profile of the data ingestion pipeline workload.
Determined to be stateless, asynchronous, queue-based, and interruption-tolerant.
Workloads capable of handling abrupt terminations benefit most from the low pricing of Spot Virtual Machines.
3
Evaluate and eliminate anti-pattern architecture proposals.
Rejected using Spot VMs for the continuous analytics engine and Cloud Functions for the memory-intensive engine.
Spot VMs introduce preemption risk to continuous services, while Cloud Functions cannot sustain long-running, memory-heavy database workloads.

Key Concept

Matching Compute Engine machine families and purchasing models (Spot VMs vs. CUDs) to workload memory demands and fault tolerance.
Rate this question

Topics

Planning Compute Engine Resources and Machine Types

More questions from this topic