gke-cost

What problem does it solve?

GKE clusters often incur unnecessary costs from overprovisioned workloads, idle node capacity, and unutilized Google Cloud cost-saving features, leading to bloated cloud spend without corresponding value.

Core Features & Use Cases

• Spot VM Configuration: Guidance for using Spot VMs for fault-tolerant workloads to cut compute costs by 60-90%, with clear suitability rules for different workload types and preemption handling best practices.
• Pod Rightsizing: Instructions for using VerticalPodAutoscaler (VPA) recommendations to eliminate overprovisioned CPU and memory requests, improving bin-packing and reducing wasted resources.
• Cost-Saving Feature Guidance: Best practices for committed use discount (CUD) purchasing, optimal machine type selection, and dev/test cluster cost optimizations to reduce spend without compromising production reliability.
• Use Case Example: A platform engineer managing a GKE Autopilot cluster with stateless batch processing workloads and production stateful services can use this skill to configure Spot VMs for batch jobs, right-size pod resource requests, and purchase 1-year CUDs for steady-state production workloads to reduce overall GKE spend by 30-50%.

Quick Start

Use the gke-cost skill to create a customized cost optimization plan for your GKE cluster that covers Spot VM setup for suitable workloads, pod rightsizing via VPA, and committed use discount purchasing guidance.