Multi-Cloud Strategy Advisor

Purpose & When-To-Use

Invoke this skill when:

• Designing a multi-cloud strategy that spans AWS, Azure, GCP, or other providers
• Evaluating vendor lock-in risks and need mitigation strategies
• Performing vendor selection with objective capability comparison
• Calculating Total Cost of Ownership (TCO) across multiple cloud platforms
• Planning hybrid cloud architectures that integrate on-premises with multiple clouds
• Designing cloud-agnostic abstractions and integration patterns
• Conducting cloud cost arbitrage analysis for workload placement

Do NOT use for single-cloud optimization (use cloud-aws-architect instead) or Kubernetes-only deployments (use cloud-native-deployment-orchestrator).

Pre-Checks

1. Time normalization: NOW_ET = 2025-10-25T21:30:36-04:00 (NIST/time.gov semantics)
2. Input validation:
   • Business requirements specify at least one: latency targets, compliance needs, or budget constraints
   • Workload characteristics include compute/storage/network demands or current infrastructure inventory
   • Risk tolerance defined: low (accept lock-in), medium (prefer portability), high (require multi-cloud)
3. Source freshness: All vendor documentation links accessed on NOW_ET; verify pricing/features current within 90 days
4. Abort conditions:
   • If requirements clearly fit single cloud provider → route to cloud-aws-architect
   • If purely Kubernetes deployment → route to cloud-native-deployment-orchestrator
   • If missing all business requirements → request clarification

Procedure

T1: Quick Multi-Cloud Assessment (≤2k tokens)

Trigger: Initial vendor selection or high-level strategy question

Steps:

1. Parse requirements to extract:
   • Compliance mandates (FedRAMP, GDPR, HIPAA) → filter vendors by region/certification
   • Latency requirements (<100ms) → identify required regions/edge locations
   • Budget constraints → tier vendors by cost profile
2. Apply decision matrix (accessed 2025-10-25T21:30:36-04:00):
   • AWS: Broadest service catalog (200+ services), strongest enterprise adoption, best for hybrid via AWS Outposts (https://aws.amazon.com/hybrid/)
   • Azure: Best Microsoft integration (AD, Office 365), strong hybrid via Azure Arc (https://learn.microsoft.com/en-us/azure/azure-arc/overview accessed 2025-10-25T21:30:36-04:00), 60+ compliance certifications
   • GCP: Best for data analytics (BigQuery), Kubernetes-native via Anthos (https://cloud.google.com/anthos accessed 2025-10-25T21:30:36-04:00), competitive ML platform (Vertex AI)
3. Generate recommendation:
   • If 1 vendor meets >90% requirements → single-cloud (cite rationale)
   • If 2-3 vendors needed → multi-cloud (proceed to T2)
   • If on-prem integration required → hybrid cloud (proceed to T2)

Output: Vendor shortlist (1-3 providers) with one-sentence justification each

T2: Multi-Cloud Strategy Design (≤6k tokens)

Trigger: Multi-cloud or hybrid architecture needed; TCO analysis required

Steps:

1. Vendor capability comparison (sources accessed 2025-10-25T21:30:36-04:00):

   • Compute: EC2 (AWS) vs Azure VMs vs Compute Engine (GCP) → scoring: flexibility, pricing, reserved instance discounts
   • Storage: S3 vs Blob Storage vs Cloud Storage → scoring: durability (11 9's), egress costs, lifecycle policies
   • Networking: VPC vs VNet vs VPC → scoring: peering options, bandwidth costs, global backbone quality
   • Serverless: Lambda vs Functions vs Cloud Functions → scoring: cold start times, concurrency limits, pricing model
   • Kubernetes: EKS vs AKS vs GKE → scoring: control plane cost ($0.10/hr AWS vs free Azure/GCP), auto-scaling, version lag
   • Hybrid: Outposts vs Arc vs Anthos → scoring: hardware ownership, management overhead, multi-cloud support

2. TCO calculation framework (https://library.educause.edu/resources/2015/4/tco-for-cloud-services-a-framework accessed 2025-10-25T21:30:36-04:00):

   • Compute costs: Instance hours × (on-demand rate OR reserved instance rate with 30-70% discount)
   • Storage costs: GB-months × storage tier rate + request costs + egress costs (major cost driver)
   • Network costs: Intra-region (free), inter-region ($0.01-0.02/GB), internet egress ($0.08-0.12/GB first 10TB)
   • Managed services: RDS/SQL Database/Cloud SQL → add 30-50% premium over self-managed
   • Hidden costs: Support (3-10% of monthly spend), training, certification, multi-cloud tooling
   • Time horizons: 3-year (standard), 5-year (enterprise), 7-year (government)
   • TCO formula: TCO = (Compute + Storage + Network + Services + Support + Training + Tooling) × Time Horizon

3. Integration pattern selection (CNCF cloud-native patterns https://www.cncf.io/ accessed 2025-10-25T21:30:36-04:00):

   • Cloud-agnostic abstraction layer:
     • Infrastructure as Code: Terraform/Pulumi for multi-cloud provisioning (not CloudFormation/ARM/Deployment Manager)
     • Orchestration: Kubernetes (EKS/AKS/GKE) for workload portability
     • Service mesh: Istio/Linkerd for cross-cloud service discovery and mTLS
   • Data integration:
     • Replication: Bidirectional (active-active) vs unidirectional (DR)
     • Event streaming: Kafka (self-managed) or Confluent Cloud (multi-cloud SaaS)
     • ETL/ELT: Airbyte, Fivetran for cross-cloud data pipelines
   • Identity federation:
     • SAML/OIDC: Single sign-on across AWS (IAM Identity Center), Azure (Entra ID), GCP (Cloud Identity)
     • Workload identity: SPIFFE/SPIRE for zero-trust service authentication
   • Network connectivity:
     • VPN: Site-to-site IPSec for low-bandwidth (<1Gbps)
     • Direct connect: AWS Direct Connect, Azure ExpressRoute, GCP Interconnect for high-bandwidth (1-100Gbps)
     • SD-WAN: Multi-cloud overlay network (Cisco Viptela, VMware VeloCloud)

4. Vendor lock-in mitigation (https://buzzclan.com/cloud/vendor-lock-in/ accessed 2025-10-25T21:30:36-04:00):

   • Data portability: Use open formats (Parquet, Avro, ORC) not proprietary (DynamoDB format, Cosmos DB)
   • API abstraction: Abstract vendor-specific APIs (S3 SDK → MinIO-compatible API)
   • Containerization: Package workloads in OCI containers (not Lambda/Functions/Cloud Functions)
   • Schema portability: Use PostgreSQL/MySQL (available on all clouds) not Aurora/CosmosDB/Spanner
   • Exit strategy: Test restore/migration quarterly; maintain cross-cloud DR every 6 months
   • Top mitigation strategies (accessed 2025-10-25T21:30:36-04:00):
     • Make informed decisions before vendor selection (66.4% organizations)
     • Maintain open environment for continuous competition (52.3%)
     • Use standard software components with proven interfaces (39.3%)

5. Generate deliverables:

   • Vendor comparison matrix: JSON with vendor, service, score (0-10), justification
   • TCO spreadsheet schema: Columns: vendor, category, unit_cost, quantity, subtotal, 3yr/5yr/7yr totals
   • Architecture diagram: Text-based (PlantUML/Mermaid) showing cloud boundaries, integration points, data flows
   • Migration roadmap: Phases (assess → pilot → migrate → optimize), duration, dependencies, rollback plan

Output: Complete multi-cloud strategy package (vendor matrix, TCO analysis, architecture, roadmap)

T3: Deep Analysis (Out of Scope for T2)

T3 tier not implemented. For deep dives:

• Cost optimization → invoke finops-cost-analyzer
• Security hardening → invoke security-assessment-framework with security-zerotrust-architect
• Kubernetes federation → invoke cloud-native-deployment-orchestrator

Decision Rules

1. Vendor selection threshold:

   • Score delta <10% → tie → select by strategic fit (existing skills, partnerships)
   • Score delta 10-30% → slight preference → validate with pilot
   • Score delta >30% → clear winner → proceed with confidence

2. Multi-cloud vs hybrid:

   • If compliance requires data residency → hybrid (on-prem + cloud)
   • If workload portability >70% → multi-cloud (2+ public clouds)
   • If vendor diversification risk mitigation → multi-cloud
   • Else → single-cloud (simplicity wins)

3. TCO ambiguity handling:

   • If egress costs >30% of total → flag for optimization (CDN, regional caching)
   • If reserved instance savings >40% → mandate 1-year commitments
   • If support costs >10% → review tier (business vs enterprise)

4. Abort conditions:

   • If user requests implementation details → stop; emit architecture and route to cloud-native-deployment-orchestrator
   • If data sovereignty conflicts with vendor regions → stop; document blockers

Output Contract

JSON Schema:

{
  "strategy_type": "multi-cloud | hybrid-cloud | single-cloud",
  "vendors": [
    {
      "name": "AWS | Azure | GCP | Other",
      "role": "primary | secondary | DR",
      "services": ["compute", "storage", "network", "managed-services"],
      "score": 0-10,
      "justification": "string (≤160 chars)"
    }
  ],
  "tco_analysis": {
    "time_horizon_years": 3 | 5 | 7,
    "total_tco_usd": "number",
    "breakdown": {
      "compute": "number",
      "storage": "number",
      "network": "number",
      "managed_services": "number",
      "support": "number",
      "training_tooling": "number"
    },
    "comparison": "string (vendor A $X vs vendor B $Y over N years)"
  },
  "integration_patterns": [
    {
      "pattern": "IaC | Kubernetes | Service Mesh | Event Streaming | Identity Federation",
      "tool": "string (Terraform, Istio, Kafka, etc.)",
      "rationale": "string (≤100 chars)"
    }
  ],
  "lock_in_mitigation": [
    {
      "risk": "data | API | schema | network",
      "mitigation": "string",
      "priority": "high | medium | low"
    }
  ],
  "roadmap": [
    {
      "phase": "assess | pilot | migrate | optimize",
      "duration_weeks": "number",
      "deliverables": ["string"],
      "dependencies": ["string"]
    }
  ],
  "sources": [
    {
      "title": "string",
      "url": "string",
      "access_date": "2025-10-25T21:30:36-04:00"
    }
  ]
}

Required fields: strategy_type, vendors (≥1), tco_analysis, integration_patterns, roadmap

Examples

# Input: Global SaaS, <50ms latency, GDPR/HIPAA, $500k/yr
requirements: {latency: <50ms, compliance: [GDPR, HIPAA],
  workloads: {vms: 100, db: 50TB, ml: GPU}, risk: medium}

# Output (T2): Multi-cloud AWS+GCP
strategy_type: multi-cloud
vendors:
  - {name: AWS, role: primary, score: 8.5,
     justification: 30 regions global}
  - {name: GCP, role: ML/analytics, score: 9.0,
     justification: Vertex AI best}
tco_5yr: $2.1M {compute: $1.2M, storage: $300k,
  network: $400k, services: $100k, support: $80k}
integration_patterns:
  - {pattern: IaC, tool: Terraform}
  - {pattern: K8s, tool: EKS+GKE}
lock_in_mitigation:
  - {risk: data, mitigation: Parquet, priority: high}
  - {risk: API, mitigation: Abstract SDKs, priority: high}
roadmap:
  - {phase: pilot, weeks: 8, deliverables: [POC AWS+GCP]}
  - {phase: migrate, weeks: 16, deliverables: [blue/green]}
sources: [aws.amazon.com/hybrid, cloud.google.com/anthos,
  cncf.io] (accessed 2025-10-25T21:30:36-04:00)

Quality Gates

1. Token budgets:

   • T1 ≤ 2k tokens (vendor shortlist only)
   • T2 ≤ 6k tokens (full strategy with TCO, architecture, roadmap)
   • T3 not implemented (route to specialized skills)

2. Safety:

   • No secrets or credentials in outputs
   • Pricing data cited with access dates (must be <90 days old)
   • Compliance claims verified against official vendor docs

3. Auditability:

   • Every TCO assumption documented (instance types, storage tiers, network estimates)
   • Every vendor score justified with comparison criteria
   • Sources include title, URL, and NOW_ET access date

4. Determinism:

   • Same inputs → same vendor scores (no randomness)
   • TCO calculations reproducible (explicit formulas)
   • Decision rules transparent and consistent

5. Abort on uncertainty:

   • If pricing data >90 days old → request manual verification
   • If workload characteristics ambiguous → request clarification
   • If compliance conflicts detected → stop and document

Resources

Official Vendor Documentation

• AWS Hybrid/Multi-Cloud: https://docs.aws.amazon.com/prescriptive-guidance/latest/strategy-education-hybrid-multicloud/introduction.html (accessed 2025-10-25T21:30:36-04:00)
• AWS Hybrid Best Practices: https://docs.aws.amazon.com/prescriptive-guidance/latest/hybrid-cloud-best-practices/introduction.html (accessed 2025-10-25T21:30:36-04:00)
• Azure Arc Overview: https://learn.microsoft.com/en-us/azure/azure-arc/overview (accessed 2025-10-25T21:30:36-04:00)
• Azure Hybrid Landing Zones: https://learn.microsoft.com/en-us/azure/cloud-adoption-framework/scenarios/hybrid/enterprise-scale-landing-zone (accessed 2025-10-25T21:30:36-04:00)
• Google Anthos: https://cloud.google.com/anthos (accessed 2025-10-25T21:30:36-04:00)

Standards & Frameworks

• CNCF Cloud Native Projects: https://www.cncf.io/ (accessed 2025-10-25T21:30:36-04:00)
• CNCF Cloud Native Maturity Model 4.0: https://www.cncf.io/blog/2025/10/22/cloud-native-maturity-model-4-0-beta-reflecting-whats-next-for-cloud-native-and-we-want-your-input/ (accessed 2025-10-25T21:30:36-04:00)
• TCO Framework (EDUCAUSE): https://library.educause.edu/resources/2015/4/tco-for-cloud-services-a-framework (accessed 2025-10-25T21:30:36-04:00)

Vendor Lock-In Mitigation

• Vendor Lock-In Prevention Guide 2025: https://buzzclan.com/cloud/vendor-lock-in/ (accessed 2025-10-25T21:30:36-04:00)
• Managing Technical Lock-In (UK Gov): https://www.gov.uk/guidance/managing-technical-lock-in-in-the-cloud (accessed 2025-10-25T21:30:36-04:00)

Multi-Cloud Patterns

• Multi-Cloud Architecture 2025: https://futransolutions.com/blog/multi-cloud-architecture-2025-the-blueprint-for-future-ready-enterprises/ (accessed 2025-10-25T21:30:36-04:00)

Resource Files (in resources/)

• vendor-comparison-matrix.json - Template for scoring AWS/Azure/GCP across service categories
• tco-calculator-template.csv - Spreadsheet template with formulas for 3/5/7 year projections
• integration-patterns.md - Common multi-cloud integration patterns with Terraform/Kubernetes examples