3.2.3.2. Capabilities of Auto Scaling for a Variety of AWS Services (EC2 Auto Scaling groups, RDS storage auto scaling, DynamoDB, ECS capacity provider, EKS autoscalers)

3.2.3.2. Auto Scaling Capabilities & Configuration

Auto Scaling ensures capacity matches demand — preventing both over-provisioning (wasted cost) and under-provisioning (poor performance).

Auto Scaling types on AWS:

EC2 ASG configuration:

# CloudFormation ASG with target tracking
AutoScalingGroup:
  Type: AWS::AutoScaling::AutoScalingGroup
  Properties:
    LaunchTemplate:
      LaunchTemplateId: !Ref LaunchTemplate
      Version: !GetAtt LaunchTemplate.LatestVersionNumber
    MinSize: 2
    MaxSize: 20
    DesiredCapacity: 4
    VPCZoneIdentifier: [!Ref SubnetA, !Ref SubnetB]
    TargetGroupARNs: [!Ref TargetGroup]
    HealthCheckType: ELB
    HealthCheckGracePeriod: 300

ScalingPolicy:
  Type: AWS::AutoScaling::ScalingPolicy
  Properties:
    AutoScalingGroupName: !Ref AutoScalingGroup
    PolicyType: TargetTrackingScaling
    TargetTrackingConfiguration:
      PredefinedMetricSpecification:
        PredefinedMetricType: ALBRequestCountPerTarget
        ResourceLabel: !Sub "${ALB.LoadBalancerFullName}/${TargetGroup.TargetGroupFullName}"
      TargetValue: 1000  # 1000 requests per target

Predictive scaling uses ML to forecast demand and pre-scale capacity. Best for workloads with predictable daily/weekly patterns.

Exam Trap: ASG DesiredCapacity overrides target tracking temporarily. If you manually set desired=10 and target tracking wants 5, the ASG stays at 10 until the next scaling evaluation. To let Auto Scaling manage capacity entirely, don't manually set DesiredCapacity after initial creation.