Autoscale an ECS Fargate service using Application Auto Scaling target tracking on CPU, memory, or ALB request count

Documented steps

1. Confirm the ECS service already exists and note its cluster and service name; Application Auto Scaling manages desired count but does not create the service.
2. Register the service as a scalable target: aws application-autoscaling register-scalable-target --service-namespace ecs --resource-id service/my-cluster/my-service --scalable-dimension ecs:service:DesiredCount --min-capacity 1 --max-capacity 10 — ResourceId must use the exact format service/<cluster>/<service>.
3. Application Auto Scaling uses the AWS-managed service-linked role AWSServiceRoleForApplicationAutoScaling_ECSService automatically for ECS; only pass --role-arn for a custom role.
4. Create a CPU-based target tracking policy: aws application-autoscaling put-scaling-policy --policy-name cpu75-target-tracking --policy-type TargetTrackingScaling --service-namespace ecs --scalable-dimension ecs:service:DesiredCount --resource-id service/my-cluster/my-service --target-tracking-scaling-policy-configuration '{"TargetValue":75.0,"PredefinedMetricSpecification":{"PredefinedMetricType":"ECSServiceAverageCPUUtilization"},"ScaleOutCooldown":60,"ScaleInCooldown":60}'
5. For memory-based scaling use PredefinedMetricType ECSServiceAverageMemoryUtilization with the same structure and TargetValue as a percentage.
6. For ALB-request-based scaling, attach the service to a target group first, then use PredefinedMetricType ALBRequestCountPerTarget and set the required ResourceLabel in the format app/<load-balancer-name>/<load-balancer-id>/targetgroup/<target-group-name>/<target-group-id>.
7. Set ScaleOutCooldown and ScaleInCooldown (seconds) inside the policy configuration to control how quickly the policy reacts again after a scaling activity.
8. Optionally set DisableScaleIn=true to allow scale-out only while keeping target tracking active.
9. To protect specific running tasks from scale-in termination: aws ecs update-task-protection --cluster my-cluster --tasks <task-id> --protection-enabled --expires-in-minutes 60 (requires ecs:UpdateTaskProtection).
10. Verify: aws application-autoscaling describe-scalable-targets --service-namespace ecs --resource-ids service/my-cluster/my-service and aws application-autoscaling describe-scaling-policies --service-namespace ecs --resource-id service/my-cluster/my-service
11. Each TargetTrackingScaling PutScalingPolicy call creates underlying CloudWatch alarms (returned in the Alarms array, named TargetTracking-...) that you should not delete manually.
12. Official docs: https://docs.aws.amazon.com/AmazonECS/latest/developerguide/service-auto-scaling.html | https://docs.aws.amazon.com/autoscaling/application/APIReference/API_RegisterScalableTarget.html | https://docs.aws.amazon.com/autoscaling/application/APIReference/API_PutScalingPolicy.html | https://docs.aws.amazon.com/autoscaling/application/APIReference/API_PredefinedMetricSpecification.html