An IoT telemetry platform is designing a Google Kubernetes Engine (GKE) cluster architecture to support two separate microservice workloads:
1. A customer-facing API service requiring high availability and operational simplicity with zero node-level infrastructure management.
2. A large-scale data transformation batch job that is fault-tolerant and stateless, where cost minimization is the top priority.
Which TWO architectural decisions should the cloud engineering team implement to satisfy these requirements? (Select TWO options.)
- Deploy GKE Autopilot mode for the customer-facing API workload to ensure Google manages node provisioning and maintenance while billing only for requested pod resources.Answer
- Configure a dedicated node pool utilizing Spot VMs for running the batch transformation workload.Answer
- CConfigure Cluster Autoscaler on the API workload to automatically scale individual Pod replicas when CPU utilization increases.
- DUtilize dedicated On-Demand Standard VM node pools with static size limits for the batch processing workloads to eliminate preemption risk and keep compute costs strictly predictable.
- EModify custom Linux kernel sysctl parameters on the Autopilot cluster nodes to optimize high-throughput socket performance for the API service.
Answer
The team should deploy GKE Autopilot mode for the API workload to eliminate node management overhead and configure a Spot VM node pool for the stateless, fault-tolerant batch job to minimize costs.
Deploying GKE Autopilot satisfies the zero node-management overhead requirement by shifting node maintenance and scaling to Google while billing only for pod requests. Provisioning a dedicated Spot VM node pool optimizes costs for stateless, interruptible batch workloads by providing heavy compute discounts.
Step-by-Step Solution
Key Concept
Planning GKE Cluster Architectures using Autopilot for managed operational boundary and Spot VMs for fault-tolerant cost reduction