Failover Architecture

active_active:
  description: "Traffic served from multiple regions simultaneously"
  benefits:
    - "Zero downtime during regional failure"
    - "Better performance (users served from nearest region)"
    - "Higher capacity (all regions handle traffic)"
  tradeoffs:
    - "Data consistency challenges"
    - "Complex replication"
    - "Higher cost (full infrastructure in each region)"

  configuration:
    regions:
      - name: "us-east-1"
        role: "active"
        weight: 50
        traffic_share: "50%"

      - name: "eu-west-1"
        role: "active"
        weight: 50
        traffic_share: "50%"

    load_balancing:
      method: "latency-based routing"
      health_check:
        interval: "10s"
        timeout: "5s"
        healthy_threshold: 2
        unhealthy_threshold: 3

    data_sync:
      method: "multi-primary replication"
      conflict_resolution: "last-write-wins"
      consistency: "eventual"

active_passive:
  description: "One region serves traffic, another is standby"
  benefits:
    - "Simpler data consistency"
    - "Lower cost than active-active"
    - "Easier to reason about"
  tradeoffs:
    - "Failover takes time"
    - "Standby infrastructure is idle"
    - "Capacity reduced during failover"

  configuration:
    regions:
      - name: "us-east-1"
        role: "active"
        traffic_share: "100%"

      - name: "eu-west-1"
        role: "passive"
        traffic_share: "0%"
        standby_type: "hot"  # or "warm" or "cold"

    failover_trigger:
      method: "health check failure"
      threshold: "3 consecutive failures"
      action: "DNS update to point to passive"

    failback:
      method: "manual"
      process: |
        1. Verify primary region is healthy
        2. Sync data from secondary to primary
        3. Verify data consistency
        4. Update DNS to point to primary
        5. Verify traffic flowing to primary
        6. Monitor for issues

dns_failover:
  provider: "Route53 / Cloudflare / Google Cloud DNS"

  configuration:
    primary_record:
      name: "api.example.com"
      type: "A"
      value: "primary-lb.example.com"
      health_check: "enabled"
      failover: "PRIMARY"

    secondary_record:
      name: "api.example.com"
      type: "A"
      value: "secondary-lb.example.com"
      health_check: "enabled"
      failover: "SECONDARY"

  health_check:
    protocol: "HTTPS"
    port: 443
    path: "/healthz"
    interval: 30
    failure_threshold: 3
    regions:
      - "us-east-1"
      - "eu-west-1"
      - "ap-southeast-1"

  ttl: 60  # Low TTL for fast failover

  considerations:
    - "DNS propagation takes time (TTL-dependent)"
    - "Some clients cache DNS longer than TTL"
    - "Health checks have their own latency"
    - "Consider using anycast for faster failover"

traffic_managers:
  global_load_balancer:
    provider: "Cloudflare Load Balancer / AWS Global Accelerator"
    benefits:
      - "Faster failover than DNS alone"
      - "Health checks at the load balancer level"
      - "Traffic steering based on latency"
      - "Automatic failover"

    configuration:
      pools:
        - name: "us-pool"
          origins:
            - address: "us-east-1.example.com"
              weight: 1
          health_checks:
            interval: 30
            timeout: 5

        - name: "eu-pool"
          origins:
            - address: "eu-west-1.example.com"
              weight: 1
          health_checks:
            interval: 30
            timeout: 5

      steering_policy:
        type: "latency"
        fallback: "random"

  kubernetes_ingress:
    provider: "NGINX Ingress / Istio / Linkerd"
    benefits:
      - "Service-level traffic management"
      - "Canary deployments"
      - "Circuit breaking"
      - "Rate limiting"

    configuration: |
      apiVersion: networking.istio.io/v1beta1
      kind: VirtualService
      metadata:
        name: user-api
      spec:
        hosts:
          - user-api
        http:
          - route:
              - destination:
                  host: user-api
                  subset: primary
                weight: 90
              - destination:
                  host: user-api
                  subset: canary
                weight: 10

regional_dependencies:
  services:
    user-api:
      primary_region: "us-east-1"
      dependencies:
        - service: "database"
          type: "regional"
          replication: "synchronous"
          failover: "automatic"

        - service: "cache"
          type: "regional"
          replication: "none"
          failover: "rebuild from database"

        - service: "payment-gateway"
          type: "global"
          replication: "n/a"
          failover: "automatic (provider managed)"

  risk_assessment:
    single_region_dependencies:
      - "If database fails, user-api fails"
      - "Mitigation: database replication to second region"

    cross_region_dependencies:
      - "If payment gateway fails, payments fail globally"
      - "Mitigation: multiple payment providers"

    global_dependencies:
      - "If DNS fails, all services unreachable"
      - "Mitigation: anycast DNS, CDN fallback"