tsh-optimizing-cloud-cost

# Cloud Cost Optimization ## When to Use - Reducing cloud spending or meeting budget constraints - Right-sizing resources based on utilization - Implementing cost governance and tagging - Analyzing multi-cloud cost optimization opportunities ## Pricing Model Decision | Workload Type | AWS | Azure | GCP | Savings | |---------------|-----|-------|-----|---------| | Steady-state (24/7) | Reserved Instances / Savings Plans | Reserved VMs | Committed Use | 30-72% | | Variable / burst | On-Demand + Auto-scaling | Pay-as-you-go | On-Demand | Baseline | | Fault-tolerant batch | Spot Instances | Spot VMs | Preemptible | Up to 90% | | Serverless / event-driven | Lambda | Functions | Cloud Functions | Pay-per-use | ## Storage Tiering Decision | Data Access Pattern | AWS | Azure | GCP | |---------------------|-----|-------|-----| | Frequent (hot) | S3 Standard | Hot | Standard | | Infrequent (30+ days) | S3 Standard-IA | Cool | Nearline | | Rare (90+ days) | S3 Glacier | Cold | Coldline | | Archive (365+ days) | S3 Deep Archive | Archive | Archive | **Rule:** Implement lifecycle policies to auto-transition data between tiers. ## Required Tags | Tag | Purpose | Example | |-----|---------|---------| | `Environment` | Cost allocation by env | `production`, `staging` | | `Project` | Cost allocation by project | `my-project` | | `CostCenter` | Finance reporting | `engineering` | | `Owner` | Contact for unused resources | `team@example.com` | | `ManagedBy` | Track IaC vs manual | `terraform` | **Reference:** See `references/tagging-standards.md` for full standards. ## Process 1. **Audit current spend** → Use Cost Explorer / Cost Management to identify top costs 2. **Tag all resources** → Apply required tags for cost allocation 3. **Identify waste** → Find unused EBS, EIPs, snapshots, idle instances 4. **Right-size resources** → Use Compute Optimizer / Advisor recommendations 5. **Apply pricing models** → Reserved/Committed for steady, Spot for batch 6. **Implement tiering** → Storage lifecycle policies, database right-sizing 7. **Set up alerts** → Budget alerts at 50%, 80%, 100% thresholds 8. **Schedule reviews** → Weekly cost review, monthly optimization ## Checklist - [ ] Cost allocation tags applied to all resources - [ ] Unused resources deleted (EBS, EIPs, snapshots, idle instances) - [ ] Instances right-sized based on utilization metrics - [ ] Reserved capacity purchased for steady workloads (>70% utilization) - [ ] Spot/preemptible used for fault-tolerant workloads - [ ] Storage lifecycle policies configured - [ ] Budget alerts enabled (50%, 80%, 100%) - [ ] Cost anomaly detection enabled - [ ] Auto-scaling configured for variable workloads ## Anti-Patterns | ❌ Don't | ✅ Do | |----------|-------| | Over-provision "just in case" | Right-size based on metrics, scale up if needed | | Use On-Demand for steady workloads | Use Reserved/Committed for predictable usage | | Store all data in hot tier | Implement lifecycle policies | | Skip tagging | Tag everything for cost visibility | | Review costs monthly | Review weekly, alert on anomalies | ## Tools | Provider | Tools | |----------|-------| | AWS | Cost Explorer, Cost Anomaly Detection, Compute Optimizer | | Azure | Cost Management, Advisor | | GCP | Cost Management, Recommender | | Multi-cloud | CloudHealth, Cloudability, Kubecost | ## Reference Files - `references/tagging-standards.md` - Tagging conventions and enforcement ## Related Skills - `tsh-implementing-terraform-modules` - For resource provisioning - `tsh-designing-multi-cloud-architecture` - For cloud selection