Multi-Cloud Strategies

---

1 - Architecture patterns

1.1 Active-Passive (DR)

Characteristics:

• RTO: 15-60 minutes
• RPO: Minutes to hours
• Cost: Moderate (minimal DR)

1.2 Active-Active

Characteristics:

• RTO: Seconds
• RPO: Near zero
• Cost: High (dual infrastructure)

1.3 Cloud Bursting

Characteristics:

• Use a secondary cloud to absorb spikes
• Cost-effective if well configured
• Requires a portable architecture

---

2 - Workload Placement

2.1 Decision criteria

Criterion	Description
Performance	Required latency, throughput
Cost	Service and egress pricing
Compliance	Regulatory requirements
Data Location	Where the data is
Features	Specific services needed
Expertise	Team's skills

2.2 Decision matrix

2.3 Placement example

Application	Cloud	Reason
API Gateway	AWS	Mature API Gateway
ML Training	GCP	TPU, Vertex AI
Office 365 Integration	Azure	Native to Microsoft
Global CDN	CloudFlare	Multi-origin
Database Primary	AWS	Mature RDS
Database DR	Azure	Provider isolation

---

3 - Workload portability

3.1 Levels of portability

Level	Description	Effort
Containers	Standard Docker	Easy
Kubernetes	K8s manifests	Medium
Terraform	Portable IaC	Medium
Managed services	Proprietary	Difficult

3.2 Recommended abstractions

# Portable architecture
Compute:
  Abstraction: Kubernetes
  Implementation:
    - AWS: EKS
    - Azure: AKS
    - GCP: GKE

Database:
  Abstraction: PostgreSQL API
  Implementation:
    - AWS: RDS PostgreSQL
    - Azure: Azure Database for PostgreSQL
    - GCP: Cloud SQL PostgreSQL

Storage:
  Abstraction: S3 API
  Implementation:
    - AWS: Native S3
    - Azure: Blob + S3 compatibility
    - GCP: Cloud Storage + S3 compatibility

Messaging:
  Abstraction: Kafka API
  Implementation:
    - AWS: MSK
    - Azure: Event Hubs
    - GCP: Managed Kafka

3.3 Avoiding Lock-in

---

4 - Multi-Cloud Disaster Recovery

4.1 DR levels

Level	RTO	RPO	Description
Backup & Restore	Hours	Hours	Cross-cloud backups
Pilot Light	Minutes	Minutes	Minimal infra active
Warm Standby	Minutes	Seconds	Reduced infra ready
Active-Active	Seconds	~0	Full redundancy

4.2 DR configuration

# dr-config.yaml
primary:
  provider: aws
  region: eu-west-1
  services:
    compute:
      type: eks
      nodes: 10
    database:
      type: rds-postgresql
      multi_az: true
    storage:
      type: s3
      replication: true

dr:
  provider: azure
  region: westeurope
  strategy: warm-standby
  services:
    compute:
      type: aks
      nodes: 2  # Scaled down
    database:
      type: azure-postgresql
      replica: async
    storage:
      type: blob
      replication: continuous

failover:
  trigger: manual  # or automatic
  dns_ttl: 60
  health_check_interval: 30s
  failover_threshold: 3

4.3 DR testing

# DR Test Checklist
1. Verify data replication
2. Validate SSL certificates
3. Test DNS failover
4. Verify network connections
5. Test authentication
6. Validate performance
7. Document the results

---

5 - Multi-Cloud Migration

5.1 Migration phases

5.2 Migration strategies

Strategy	Description	Use Case
Lift & Shift	Move as-is	Fast migration
Replatform	Minor adjustments	Containers
Refactor	Rearchitecture	Cloud-native
Repurchase	New service	SaaS
Retire	Remove	Unnecessary legacy

---

Summary

In this chapter, we learned:

• The architecture patterns (Active-Passive, Active-Active)
• Workload placement
• Workload portability
• Disaster Recovery strategies
• Migration approaches

---

Next step

In the next chapter, we will look at Terraform Multi-Cloud.

→ Next chapter: Terraform Multi-Cloud

---

← Back to table of contents