Examkin

Question

Difficulty: Very hardPlanning Google Kubernetes Engine (GKE) Cluster Architectures

A financial technology company is designing a container architecture on Google Cloud for two distinct services. The first service is a stateless public API frontend that requires minimal operational overhead, automated node security patching, and no infrastructure management. The second service is a fault-tolerant batch processing engine that executes asynchronous calculations, where workload interruptions are acceptable in exchange for maximum cost savings. Which TWO architectural decisions should the cloud engineering team implement to satisfy these requirements?

  1. Deploy a GKE Autopilot cluster to host the stateless public API frontend workload.Answer
  2. Provision a GKE Standard node pool utilizing Spot VMs with Cluster Autoscaler enabled for the batch processing engine.Answer
  3. C
    Deploy a GKE Standard cluster with custom-managed worker node pools for the API frontend to eliminate node-level operational management.
  4. D
    Utilize Spot VM node pools to host the application's primary persistent relational database tier to reduce infrastructure expenses.
  5. E
    Configure the Horizontal Pod Autoscaler (HPA) to dynamically increase the Compute Engine node count from zero when batch jobs are submitted.

Answer

The team should deploy a GKE Autopilot cluster for the stateless API frontend and configure a GKE Standard node pool using Spot VMs with Cluster Autoscaler for the batch processing engine.
Choosing GKE Autopilot transfers all node management, OS patching, and security provisioning duties directly to Google Cloud, satisfying the requirement for zero infrastructure management overhead for the stateless API frontend. For the fault-tolerant batch processing engine, establishing a GKE Standard node pool composed of Spot VMs provides significant cost reduction for workloads that tolerate preemption, while Cluster Autoscaler ensures nodes scale dynamically based on pending pod queue requirements.

Step-by-Step Solution

1
Analyze operational requirements for the stateless API frontend
Identified the requirement for zero node management overhead and automated security patching.
GKE Autopilot manages the entire node lifecycle and infrastructure, making it the ideal choice when node management is to be eliminated.
2
Analyze cost and resiliency requirements for the batch processing engine
Identified a fault-tolerant, interruptible workload requiring maximum cost reduction.
Spot VMs offer up to 60-91% discounts over standard VMs, perfect for stateless batch jobs where node preemption is acceptable.
3
Select infrastructure auto-provisioning strategy for node capacity
Paired Spot VMs in GKE Standard with Cluster Autoscaler.
Cluster Autoscaler automatically adjusts node count based on resource demands of pending batch pods, whereas HPA only adjusts pod counts.

Key Concept

Planning Google Kubernetes Engine (GKE) Cluster Architectures: GKE Autopilot vs. Standard Operational Boundaries and Spot VM Integration
Rate this question

Topics

Planning Google Kubernetes Engine (GKE) Cluster Architectures

More questions from this topic