Skip to main content

Cost Optimization

Status

Master Plan - Comprehensive init.md complete, ready for SKILL.md implementation

Comprehensive FinOps practices for cloud cost optimization, covering visibility, commitment discounts, right-sizing, and Kubernetes cost management.

Scope

This skill provides FinOps frameworks and automation:

  • Cost Visibility - Tagging, showback/chargeback, anomaly detection
  • Commitment Discounts - Reserved Instances, Savings Plans, Committed Use Discounts
  • Right-Sizing - Compute, database, storage, Kubernetes optimization
  • Spot Instances - 70-90% savings for fault-tolerant workloads
  • Kubernetes Cost Management - Kubecost, OpenCost, resource requests/limits
  • Automation - Idle resource cleanup, budget alerts, policy enforcement

Key Components

  • FinOps Lifecycle - Inform → Optimize → Operate (continuous improvement)
  • Cost Allocation Tags - Enable showback/chargeback reporting
  • Commitment Strategy - When to purchase Reserved Instances vs. Savings Plans
  • Kubernetes Rightsizing - VPA recommendations, namespace quotas, cluster autoscaling
  • Budget Alerts - Cascading notifications at 50%, 75%, 90%, 100%
  • Unit Cost Metrics - Cost per customer, cost per transaction (business value)

Decision Framework

Commitment Discount Strategy:

Steady-state baseline (6+ months) → Reserved Instances (40-72% savings)
Variable compute workload → Savings Plans (flexible)
Fault-tolerant batch jobs → Spot Instances (70-90% savings)
Unpredictable/spiky traffic → On-Demand (no commitment)

Right-Sizing Priority:

HIGH: Idle resources (100% waste) - Delete unattached volumes, stopped instances
MEDIUM: Over-provisioned instances (<40% utilization) - Downsize
LOW: Application optimization - Requires code changes

Tool Recommendations

Cost Visibility:

  • Kubecost - Kubernetes cost allocation and optimization
  • Infracost - Terraform cost estimation in CI/CD
  • CloudZero - Unit cost economics and anomaly detection
  • AWS Cost Explorer / Azure Cost Management / GCP Cloud Billing - Native cloud tools

Automation:

  • AWS Compute Optimizer - ML-based EC2 rightsizing
  • Azure Advisor - Multi-service optimization recommendations
  • GCP Recommender - Idle resource detection, commitment advice

Integration Points

  • resource-tagging - Cost allocation tags enable showback/chargeback
  • operating-kubernetes - K8s rightsizing, VPA, cluster autoscaling
  • writing-infrastructure-code - Infracost for Terraform cost estimation
  • deploying-on-aws / deploying-on-gcp / deploying-on-azure - Cloud-specific optimizations
  • platform-engineering - Internal FinOps platforms and self-service tools

Learn More