Examkin

Question

Difficulty: MediumPlanning Google Kubernetes Engine (GKE) Cluster Architectures

A smart grid utility provider is planning a Google Kubernetes Engine (GKE) cluster architecture to handle real-time telemetry from millions of smart meters. The deployment requires two distinct workloads:

1. A high-throughput API gateway service that requires custom Linux kernel (`sysctl`) parameter modifications on the node host to optimize TCP socket buffers.
2. A stateless, fault-tolerant batch data processing pipeline that parses incoming telemetry logs and can handle sudden node preemptions.

The architecture team wants to minimize compute costs and management overhead while meeting all workload technical requirements. Which TWO cluster planning choices should be included in the solution architecture?

  1. Provision a GKE Standard cluster with a node pool configured with custom Linux kernel settings to host the API gateway service.Answer
  2. Create a dedicated node pool using Spot VMs for the batch data processing pipeline workload.Answer
  3. C
    Deploy the API gateway service to a GKE Autopilot cluster to eliminate node management overhead.
  4. D
    Configure Spot VMs for the node pool hosting the API gateway service to minimize baseline infrastructure costs.
  5. E
    Configure the Horizontal Pod Autoscaler (HPA) to dynamically increase worker node VM instances when batch queue lengths spike.

Answer

The architecture plan should include provisioning a GKE Standard cluster with a node pool configured for custom kernel parameters for the API gateway service, and creating a dedicated Spot VM node pool for the stateless batch processing pipeline.
GKE Standard must be selected because the API gateway requires custom Linux kernel (`sysctl`) tuning, which is restricted in GKE Autopilot. Furthermore, utilizing Spot VMs for the batch data processing pipeline effectively minimizes compute costs, as the workload is stateless and resilient to node terminations.

Step-by-Step Solution

1
Determine the appropriate GKE cluster mode based on host kernel requirements.
Select GKE Standard for the API gateway because GKE Autopilot restricts host-level kernel parameter (sysctl) customizations.
Autopilot operates under a strict security and configuration model that blocks custom node kernel tuning.
2
Determine the optimal compute purchasing model for the batch processing pipeline.
Assign Spot VMs to the batch processing node pool to minimize infrastructure spend.
Stateless, fault-tolerant workloads can easily recover from preemption, making them ideal candidates for discounted Spot instances.

Key Concept

Evaluating GKE operational mode boundaries (Standard vs. Autopilot) and node pool provisioning models (Standard vs. Spot VMs) based on workload requirements.
Rate this question

Topics

Planning Google Kubernetes Engine (GKE) Cluster Architectures

More questions from this topic