Fondamentaux du DR et de la HA

RTO et RPO

Deux metriques fondamentales definissent votre strategie de reprise :

• RTO (Recovery Time Objective) : duree maximale acceptable d'interruption du service. Combien de temps pouvez-vous etre hors ligne ?
• RPO (Recovery Point Objective) : quantite maximale de donnees que vous pouvez perdre. Quel age peut avoir votre derniere sauvegarde ?

Strategies de sauvegarde

# Velero pour sauvegarder Kubernetes
# Installation
velero install \
  --provider aws \
  --bucket velero-backups \
  --secret-file ./credentials-velero

# Sauvegarde planifiee
velero schedule create daily-backup \
  --schedule="0 2 * * *" \
  --include-namespaces production \
  --ttl 720h

# Restauration
velero restore create --from-backup daily-backup-20260601

# Sauvegarde base de donnees avec Terraform
resource "aws_db_instance" "main" {
  identifier     = "prod-db"
  engine         = "postgres"
  engine_version = "15"

  backup_retention_period = 7
  backup_window          = "03:00-04:00"

  # Replique de lecture cross-region
  replicate_source_db = aws_db_instance.main.arn
}

Architecture multi-region

# Terraform : infrastructure multi-region
provider "aws" {
  alias  = "primary"
  region = "eu-west-1"
}

provider "aws" {
  alias  = "secondary"
  region = "eu-west-3"
}

# Route53 Health Check + Failover
resource "aws_route53_health_check" "primary" {
  fqdn              = "primary.example.com"
  port              = 443
  type              = "HTTPS"
  failure_threshold = 3
  request_interval  = 10
}

resource "aws_route53_record" "app" {
  zone_id = aws_route53_zone.main.zone_id
  name    = "app.example.com"
  type    = "A"

  failover_routing_policy {
    type = "PRIMARY"
  }

  set_identifier  = "primary"
  health_check_id = aws_route53_health_check.primary.id

  alias {
    name    = aws_lb.primary.dns_name
    zone_id = aws_lb.primary.zone_id
  }
}

Failover Kubernetes

# Federation de clusters avec Submariner ou Liqo
# PodDisruptionBudget pour la HA
apiVersion: policy/v1
kind: PodDisruptionBudget
metadata:
  name: mon-app-pdb
spec:
  minAvailable: 2
  selector:
    matchLabels:
      app: mon-app

# topologySpreadConstraints pour repartir les Pods
spec:
  topologySpreadConstraints:
  - maxSkew: 1
    topologyKey: topology.kubernetes.io/zone
    whenUnsatisfiable: DoNotSchedule
    labelSelector:
      matchLabels:
        app: mon-app