Weight: 15% | Focus: High Availability, Auto Scaling, Load Balancing, Multi-AZ

🔥 MUST KNOW Services

Auto Scaling - Dynamic Capacity

  • ASG: EC2 instances across AZs
  • Scaling Policies: Target Tracking, Step, Simple
  • Health Checks: EC2, ELB
  • Lifecycle Hooks: Custom launch/terminate actions

Elastic Load Balancer - Traffic Distribution

  • ALB: Layer 7 (HTTP/HTTPS), path/host routing
  • NLB: Layer 4 (TCP/UDP), static IPs, high performance
  • GWLB: Layer 3, third-party appliances
  • Health Checks: Determine target availability

RDS Multi-AZ - Database HA

  • Multi-AZ: Synchronous replication, automatic failover
  • Read Replicas: Asynchronous, read scaling
  • Cross-Region: Disaster recovery
  • Failover: 60-120 seconds

Route 53 - DNS & Global Load Balancing

  • Routing Policies: Simple, Weighted, Latency, Failover, Geolocation
  • Health Checks: HTTP/HTTPS/TCP monitoring
  • Alias Records: Free, automatic IP resolution

S3 Cross-Region Replication - Data Resilience

  • CRR: Cross-region disaster recovery
  • SRR: Same-region compliance
  • RTC: 15-minute replication SLA

⚡ Key Patterns

Multi-AZ Web Application

Route 53 → CloudFront → ALB (Multi-AZ) → ASG (Multi-AZ) → RDS Multi-AZ

Auto Scaling Policies

  • Target Tracking: Maintain specific metric value
  • Step Scaling: Scale based on metric ranges
  • Predictive: ML-based forecasting

RDS Failover Scenarios

  • Primary DB failure → Automatic failover to standby
  • AZ outage → Failover within 60-120 seconds
  • Maintenance → Zero-downtime updates

🎯 Exam Scenarios

Scenario 1: Application not scaling properly

  • Check Auto Scaling policies and thresholds
  • Verify CloudWatch metrics are being published
  • Review health check configuration
  • Check cooldown periods

Scenario 2: Database failover taking too long

  • Verify Multi-AZ is enabled
  • Check application connection string (use endpoint, not IP)
  • Review security group rules
  • Ensure proper error handling in application

Scenario 3: Global application latency issues

  • Implement Route 53 latency-based routing
  • Use CloudFront for static content
  • Deploy read replicas in multiple regions
  • Consider Aurora Global Database

📋 Quick Commands

# Auto Scaling
aws autoscaling create-auto-scaling-group --auto-scaling-group-name MyASG
aws autoscaling put-scaling-policy --policy-name ScaleUp --auto-scaling-group-name MyASG

# Load Balancer
aws elbv2 create-load-balancer --name MyALB --subnets subnet-12345 subnet-67890
aws elbv2 create-target-group --name MyTargets --protocol HTTP --port 80

# RDS
aws rds create-db-instance --db-instance-identifier MyDB --multi-az
aws rds reboot-db-instance --db-instance-identifier MyDB --force-failover

# Route 53
aws route53 create-health-check --type HTTP --resource-path /health
aws route53 change-resource-record-sets --hosted-zone-id Z123 --change-batch file://change.json

# S3 Replication
aws s3api put-bucket-replication --bucket MyBucket --replication-configuration file://replication.json

🏗️ Resilience Design Principles

High Availability

  • Deploy across multiple AZs
  • Use load balancers for distribution
  • Implement health checks
  • Plan for component failures

Fault Tolerance

  • Design for graceful degradation
  • Implement circuit breakers
  • Use retry mechanisms with backoff
  • Isolate failure domains

Disaster Recovery

  • RTO: Recovery Time Objective
  • RPO: Recovery Point Objective
  • Strategies: Backup/Restore, Pilot Light, Warm Standby, Multi-Site

🔧 Auto Scaling Best Practices

  • Use target tracking for most scenarios
  • Set appropriate health check grace periods
  • Implement lifecycle hooks for stateful apps
  • Monitor scaling activities and metrics
  • Use mixed instance types for cost optimization

⚠️ Common Mistakes

  • Not enabling Multi-AZ for production databases
  • Using IP addresses instead of DNS names
  • Insufficient health check grace periods
  • Not testing failover procedures
  • Forgetting cross-zone load balancing
  • Not implementing proper retry logic in applications

🔑 Key Points

  • Multi-AZ ≠ Read Replicas (HA vs Scaling)
  • ALB supports path/host routing, NLB for performance
  • Auto Scaling maintains desired capacity across AZs
  • Route 53 health checks enable DNS failover
  • RDS failover is automatic with Multi-AZ
  • S3 replication requires versioning enabled