Examkin

Question

Difficulty: HardPlanning Compute Engine Resources and Machine Types

A cloud engineer is planning the Google Compute Engine architecture for a specialized genomics research platform that consists of two distinct workloads:

1. Workload 1: A mission-critical, stateful primary database service running continuous 24/7 operations with steady, predictable memory and CPU utilization.
2. Workload 2: A batch alignment pipeline that processes queue-based genomic data blocks. The pipeline is stateless, fully fault-tolerant, saves progress checkpoints to Cloud Storage every 5 minutes, and can handle abrupt node terminations.

Which TWO deployment and resource planning choices should the engineer recommend to satisfy operational requirements while optimizing total cost of ownership? (Select TWO.)

  1. Provision standard Compute Engine VM instances backed by Committed Use Discounts (CUDs) for Workload 1.Answer
  2. Provision Spot VM instances within Managed Instance Groups (MIGs) for Workload 2.Answer
  3. C
    Provision Spot VM instances for Workload 1 to achieve maximum baseline infrastructure savings.
  4. D
    Rely exclusively on automatic Sustained Use Discounts (SUDs) for Workload 2 batch processing jobs by running them on a single large persistent instance.

Answer

The optimal solution is to provision standard Compute Engine instances with Committed Use Discounts for the continuous stateful primary database (Workload 1) and to provision Spot VM instances within Managed Instance Groups for the fault-tolerant batch alignment pipeline (Workload 2).
The combination of standard instances backed by Committed Use Discounts for the continuous stateful database workload and Spot VMs inside Managed Instance Groups for the stateless batch processing pipeline aligns perfectly with Google Cloud compute optimization best practices.

Step-by-Step Solution

1
Analyze operational characteristics of Workload 1
Identified Workload 1 as a mission-critical 24/7 stateful database needing high availability and predictable continuous compute.
Stateful 24/7 workloads require standard non-preemptible VM instances. Applying 1-year or 3-year Committed Use Discounts (CUDs) offers substantial pricing discounts for predictable baseline resource needs.
2
Analyze operational characteristics of Workload 2
Identified Workload 2 as a stateless, fault-tolerant batch workload capable of resuming from Cloud Storage checkpoints.
Fault-tolerant batch jobs that process queue-based work items can tolerate sudden instance preemptions, making Spot VMs in Managed Instance Groups the best choice to achieve up to 60-91% cost reduction.
3
Evaluate and reject anti-pattern distractors
Rejected using Spot VMs for stateful databases and relying on Sustained Use Discounts on a single large instance for batch jobs.
Spot VMs violate SLA requirements for stateful continuous workloads, while Sustained Use Discounts do not provide the extreme cost savings offered by Spot instances for checkpointed batch operations.

Key Concept

Compute Engine Resource Planning: Spot VMs vs. Committed Use Discounts (CUDs)
Rate this question

Topics

Planning Compute Engine Resources and Machine Types

More questions from this topic