Multi-Cloud FinOps Strategy

Multiple cloud providers mean multiple billing systems, discount programs, pricing models, and cost allocation methodologies. Without a unified FinOps strategy, multi-cloud becomes multi-cost — each team optimizes their own cloud while the organization’s total spend grows unchecked.

---

Cross-Cloud Cost Normalization

Problem: Comparing costs across clouds is like comparing currencies.

AWS:         m6i.xlarge  =  4 vCPU, 16 GB RAM  =  $0.192/hour
Google Cloud: n2-standard-4 = 4 vCPU, 16 GB RAM  =  $0.194/hour
Azure:       D4s v5      =  4 vCPU, 16 GB RAM  =  $0.192/hour

BUT: Same spec ≠ same performance
  CPU: Different generations, clock speeds
  Network: Different bandwidth limits
  Storage: Different IOPS tiers included
  
Normalization approach:
  Cost per unit of work (not cost per hour)
  
  Metric: Cost per 1 million API requests processed
  AWS:    $12.50 (Lambda + API Gateway)
  GCP:    $10.80 (Cloud Functions + Cloud Endpoints)
  Azure:  $11.20 (Functions + API Management)
  
  Metric: Cost per TB stored per month
  AWS S3:      $23.00
  GCS:         $20.00
  Azure Blob:  $18.40

---

Unified Tagging Strategy

# Cross-cloud tagging standard
tagging_policy:
  required_tags:
    - key: "cost-center"
      description: "Business unit cost center code"
      format: "CC-[0-9]{4}"
      example: "CC-4521"
      
    - key: "environment"
      description: "Deployment environment"
      values: ["production", "staging", "development", "sandbox"]
      
    - key: "service"
      description: "Application or service name"
      format: "lowercase-kebab-case"
      example: "order-processing"
      
    - key: "owner"
      description: "Team responsible for the resource"
      format: "team-name"
      example: "platform-team"
      
    - key: "cloud"
      description: "Cloud provider (for unified reporting)"
      values: ["aws", "gcp", "azure"]

  enforcement:
    aws: "AWS Config rule + SCP deny untagged"
    gcp: "Organization Policy + IAM conditions"
    azure: "Azure Policy + deny effect"
    
  reporting:
    tool: "Apptio, CloudHealth, or custom dashboard"
    frequency: "Weekly cost report by cost-center + service"
    anomaly: "Alert on >15% week-over-week increase"

---

Discount Portfolio

class MultiCloudDiscountManager:
    """Manage discount programs across cloud providers."""
    
    portfolio = {
        "aws": {
            "savings_plans": {
                "type": "Compute Savings Plan",
                "commitment": "$50/hour",
                "term": "1 year",
                "savings": "30%",
                "coverage": "EC2, Lambda, Fargate",
            },
            "reserved_instances": {
                "type": "Standard RI",
                "instances": "50x m6i.xlarge",
                "term": "1 year, partial upfront",
                "savings": "40%",
            },
        },
        "gcp": {
            "committed_use": {
                "type": "CUD (Committed Use Discount)",
                "resource": "200 vCPU + 800 GB RAM",
                "term": "1 year",
                "savings": "37%",
            },
            "sustained_use": {
                "type": "SUD (automatic)",
                "threshold": "25%+ monthly usage",
                "savings": "Up to 30% automatic",
            },
        },
        "azure": {
            "reserved_instances": {
                "type": "Azure RI",
                "instances": "30x D4s_v5",
                "term": "1 year",
                "savings": "35%",
            },
        },
    }
    
    def total_savings(self):
        total_on_demand = sum(p["on_demand_equivalent"] for p in self.portfolio.values())
        total_actual = sum(p["actual_spend"] for p in self.portfolio.values())
        return {
            "on_demand_equivalent": total_on_demand,
            "actual_spend": total_actual,
            "total_savings": total_on_demand - total_actual,
            "savings_rate": (total_on_demand - total_actual) / total_on_demand,
        }

---

Anti-Patterns

Anti-Pattern	Consequence	Fix
Separate FinOps per cloud	No unified cost view	Single FinOps platform across all clouds
Inconsistent tagging	Cannot compare costs cross-cloud	Unified tagging policy enforced on all providers
Optimize each cloud independently	Miss opportunities for workload placement	Total cost optimization including egress
No cross-cloud benchmarking	Overpay without knowing it	Regular pricing comparison per workload type
One team for all clouds	Expertise too thin	Cloud-specific expertise with unified FinOps governance

Multi-cloud FinOps is harder than single-cloud FinOps — but the same principles apply: visibility, optimization, and governance. The key addition is normalization: creating a common language for cost that spans providers and lets you compare apples to apples.