Providers Comparison — 🔍 Detailed Service Mapping and Architecture Patterns

Service Equivalency Matrix

Compute Services (Detailed)

Performance Characteristics:

Lambda cold start: 200-1000ms
Cloud Functions cold start: 100-600ms
Azure Functions cold start: 150-800ms

Selection: GCP/Azure fastest for latency-sensitive FaaS

Storage Services (Detailed)

Database Services (Detailed)

SQL Databases:

NoSQL Databases:

Pricing Deep Dive

AWS EC2 (t3.medium in us-east-1)

Option 1: On-demand
$0.0416/hour × 730 hours/month = $30.37/month

Option 2: 1-year reserved (33% discount)
$0.0278/hour × 730 = $20.29/month

Option 3: 3-year reserved (38% discount)
$0.0258/hour × 730 = $18.83/month

Option 4: Spot instances (70% discount)
$0.0125/hour × 730 = $9.13/month (can be interrupted)

Best for continuous workload: 3-year reserved
Best for variable workload: Spot with fallback

GCP Compute Engine (e2-medium in us-central1)

Option 1: On-demand
$0.0336/hour × 730 = $24.53/month

Option 2: 1-year commitment (25% discount)
$0.0252/hour × 730 = $18.40/month

Option 3: Preemptible (70% discount)
$0.0101/hour × 730 = $7.37/month

Automatic sustained-use discount: 10-30% (auto-applied)
Result: On-demand effectively becomes $21.59/month

Azure Virtual Machines (B2s in eastus)

Option 1: Pay-as-you-go
$0.0564/hour × 730 = $41.17/month

Option 2: 1-year reserved (40% discount)
$0.0338/hour × 730 = $24.67/month

Option 3: 3-year reserved (55% discount)
$0.0254/hour × 730 = $18.54/month

Option 4: Spot (90% discount)
$0.0056/hour × 730 = $4.09/month

Total Cost Comparison (12-month commitment)

AWS (3-year reserved):     $18.83/month
GCP (1-year commitment):   $18.40/month
Azure (3-year reserved):   $18.54/month

Winner: GCP (slightly cheaper, easier to get discount)
Variance: <3% (not significant)

Feature Comparison: Kubernetes (Container Orchestration)

Total Cost (100-node cluster, 1 year reserved):

AWS EKS:
├─ Master: $0.10/hour × 730 = $73/month
├─ 100 nodes × $18.83 = $1,883/month
└─ Total: $1,956/month

GCP GKE:
├─ Master: Free
├─ 100 nodes × $18.40 = $1,840/month
└─ Total: $1,840/month (11% cheaper)

Azure AKS:
├─ Master: Free
├─ 100 nodes × $18.54 = $1,854/month
└─ Total: $1,854/month (5% cheaper than AWS)

Winner: GCP (no master cost) or Azure (slightly cheaper nodes)

Regional Availability and Compliance

Data Residency Requirements

GCP Regions Available:

Americas: US (3 regions)
Europe: Netherlands, Belgium, Finland, Germany
Asia-Pacific: Japan, Taiwan, Singapore, Australia, India
Total: 42 zones across 14 regions

Azure Regions Available:

Americas: US (5 regions), Canada, Brazil
Europe: UK, France, Germany, Switzerland, Netherlands
Asia-Pacific: Japan (2), South Korea, India
Special: Government (GovCloud), China (special agreement)
Total: 60+ regions

AWS Regions Available:

Americas: US (7 regions), Canada, Brazil, Mexico
Europe: UK, France, Germany, Sweden
Asia-Pacific: Singapore, Japan (2), South Korea, Australia, India
Special: GovCloud, China (partition)
Total: 33 regions

Implication: Azure > AWS > GCP for global coverage

Machine Learning and Data Services

AWS:

• SageMaker (general ML)

• Rekognition (computer vision)

• Forecast (time series)

GCP:

• Vertex AI (unified ML)

• Vision AI (computer vision, best-in-class)

• BigQuery ML (SQL-based ML)

Azure:

• Azure ML (enterprise ML)

• Custom Vision (vision)

• Cognitive Services (NLP, translation)

Winner for ML: GCP (best Vision AI, BigQuery ML) if using structured data; AWS if enterprise need; Azure for enterprise Microsoft ecosystem

Organizational Decision Framework

Choose AWS if:

• Existing AWS investment

• Widest service selection needed

• Enterprise support critical

• ECS container strategy

Choose GCP if:

• Cost optimization critical

• Machine learning focus

• Data/analytics emphasis

• Kubernetes-first approach

• Best-in-class BigQuery

Choose Azure if:

• Microsoft ecosystem (Windows, SQL Server, Office 365)

• Enterprise compliance strict

• Direct on-premise integration needed

• .NET/Windows developer base

Migration Complexity by Service

Easier to migrate (standard tech):

• PostgreSQL databases (standard SQL)

• Docker containers (OCI standard)

• Python/Node applications

Harder to migrate (proprietary):

• AWS Lambda → GCP Cloud Functions (code changes)

• DynamoDB → Firestore (data model different)

• RDS MySQL → Cloud SQL (replication needed)

• S3 → Cloud Storage (API differences)

Recommendation: Start with standard services, add proprietary where advantageous

Key Takeaways

• ✅ Pricing within 5% across major providers for standard services

• ✅ GCP best for data/ML workloads

• ✅ AWS most mature for enterprise

• ✅ Azure best for Microsoft integration

• ✅ Kubernetes costs favor GCP/Azure (free master)

• ✅ Regional availability highest in Azure

• ✅ Avoid proprietary services in multi-cloud strategy

• ✅ Lock-in is real; plan for it