A fintech organization is designing a Google Kubernetes Engine (GKE) architecture to run two distinct workloads in a single production environment:
1. A high-throughput, latency-sensitive payment routing service requiring custom Linux kernel parameters (`sysctl` settings) and dedicated node CPU/memory reservations.
2. An asynchronous financial model simulation batch workload that handles transient, fault-tolerant calculation jobs with minimal cost.
Which two cluster design choices should the cloud engineer implement to meet these technical and financial requirements?
- Provision a GKE Standard cluster to enable custom Linux kernel sysctl configurations on node pools.Answer
- Configure a dedicated node pool utilizing Spot VMs for the asynchronous financial model simulation batch workload.Answer
- CProvision a GKE Autopilot cluster because it completely eliminates node management overhead while permitting custom node sysctl modifications.
- DConfigure the latency-sensitive payment routing workload to run on Spot VM node pools to minimize infrastructure costs.
- EDeploy the payment routing workload in a GKE Autopilot cluster and use Kubernetes Horizontal Pod Autoscaler to adjust physical node OS kernel parameters dynamically.
Answer
Select a GKE Standard cluster to support custom sysctl kernel configurations and use a dedicated Spot VM node pool for the fault-tolerant batch simulation workload.
Choosing GKE Standard fulfills the strict requirement for custom sysctl kernel configurations, which Autopilot prevents. Utilizing Spot VMs for the batch simulation workload satisfies the cost optimization requirement while ensuring compute preemption does not risk data integrity, as the job is fault-tolerant.
Step-by-Step Solution
Key Concept
Selecting between GKE Standard and Autopilot based on node customization needs, and leveraging Spot VMs for fault-tolerant workloads.