Examkin

Question

Difficulty: MediumPlanning Compute Engine Resources and Machine Types

A financial analytics company is designing the compute architecture on Google Compute Engine for two distinct backend services:

1. Batch Risk Simulation Engine: A highly fault-tolerant, stateless workload that processes data chunks independently and can handle sudden VM preemption without losing progress.
2. In-Memory Ledger Cache: A dedicated 24/7 database service that requires a custom ratio of 6 vCPUs and 45 GB of RAM to avoid paying for unused vCPUs.

Which TWO architectural decisions should the cloud engineer select to optimize both performance and cost for these workloads?

  1. Provision Spot VMs (or Preemptible VMs) for the Batch Risk Simulation Engine to achieve significant cost savings.Answer
  2. Configure a Custom Machine Type with extended memory for the In-Memory Ledger Cache to match the exact vCPU and RAM requirements.Answer
  3. C
    Provision Spot VMs for the In-Memory Ledger Cache to minimize baseline hourly costs.
  4. D
    Deploy the Batch Risk Simulation Engine on Cloud Functions to automatically manage scaling without using virtual machines.

Answer

The correct decisions are to use Spot VMs for the fault-tolerant batch risk simulation engine and to configure a Custom Machine Type with extended memory for the 24/7 in-memory cache.
Selecting Spot VMs for the batch simulation engine takes advantage of discount pricing suitable for fault-tolerant workloads. Configuring a Custom Machine Type for the cache allows exact allocation of 6 vCPUs and 45 GB RAM without paying for unused capacity in standard predefined shapes.

Step-by-Step Solution

1
Analyze Workload 1 (Batch Risk Simulation Engine)
Identified as fault-tolerant, stateless, and batch-oriented.
Fault-tolerant batch jobs benefit from Spot VMs because preemption does not cause data loss, providing up to 60-91% cost savings.
2
Analyze Workload 2 (In-Memory Ledger Cache)
Identified as a continuous 24/7 stateful workload requiring non-standard core/RAM ratio (6 vCPUs, 45 GB RAM).
Custom machine types let you select exact vCPU counts and memory amounts, avoiding the cost of larger standard predefined tiers.
3
Evaluate Distractor Options
Eliminated using Spot VMs for stateful 24/7 caches and using Cloud Functions for heavy batch jobs.
Spot VMs risk unexpected downtime for databases/caches, and Cloud Functions has execution timeout and resource limits unsuited for heavy simulations.

Key Concept

Compute Engine Machine Type Selection and Spot VM Suitability
Estimated Time:1m 30s
Rate this question

Topics

Planning Compute Engine Resources and Machine Types

More questions from this topic