Examkin

Question

Difficulty: HardPlanning and Assessing Cloud Compute Solutions

An enterprise retail organization is planning the migration of its order fulfillment processing system to Google Cloud. The workload comprises two core services with distinct operational profiles:

1. Fulfillment API: A stateless HTTP web service packaged as a main application container alongside a sidecar telemetry container. It receives unpredictable burst traffic during flash sales events and experiences long idle periods, requiring automatic scaling down to zero instances to eliminate costs when idle.
2. Inventory Reconciliation Worker: An asynchronous, fault-tolerant batch processing application that runs overnight to reconcile inventory databases. Individual worker tasks are stateless and designed to checkpoint progress so interrupted tasks can resume safely without data loss.

Which TWO compute platform deployment strategies should you recommend to satisfy these requirements while optimizing cost and minimizing operational overhead?

  1. Deploy the Fulfillment API to Cloud Run using multi-container deployment support.Answer
  2. B
    Deploy the Fulfillment API as a collection of Cloud Functions single-event snippets.
  3. C
    Deploy the Inventory Reconciliation Worker on GKE Autopilot clusters requiring custom OS kernel sysctl tuning parameters on node pools.
  4. Deploy the Inventory Reconciliation Worker on Compute Engine Spot VM instances managed by a Managed Instance Group (MIG).Answer
  5. E
    Deploy the Inventory Reconciliation Worker on Compute Engine Spot VM instances relying on local boot disks to store persistent state across preemptions.

Answer

The optimal deployment strategies are deploying the multi-container HTTP API service to Cloud Run and executing the fault-tolerant batch inventory worker on Compute Engine Spot VMs managed within a Managed Instance Group (MIG).
Cloud Run is the ideal choice for stateless containerized web applications needing multi-container (sidecar) support and automatic scaling down to zero when idle. For batch workloads that are fault-tolerant and implement state checkpointing, Compute Engine Spot VMs in a Managed Instance Group deliver massive cost savings with built-in resilience against preemption.

Step-by-Step Solution

1
Assess compute platform options for the HTTP API service with sidecar logging.
Identify that Cloud Run supports multi-container instances (sidecars) and scales to zero when traffic stops, matching the stateless bursty HTTP API requirements.
Cloud Functions does not support multi-container sidecar deployments, whereas GKE would incur continuous cluster management overhead and baseline node costs during idle periods.
2
Assess compute options for the fault-tolerant nightly batch workload.
Select Compute Engine Spot VMs in a Managed Instance Group (MIG).
Because the batch application is stateless, fault-tolerant, and checkpoints its state, taking advantage of Spot VM discounts offers maximum cost savings without risk of data loss.

Key Concept

Selecting GCP Compute Services Based on Workload Characteristics and Cost Trade-offs
Estimated Time:2m 0s
Rate this question

Topics

Planning and Assessing Cloud Compute Solutions

More questions from this topic