Service Auto Scaling

Automatically adjust the desired count using Application Auto Scaling policies.

Service auto scaling changes the desired count in response to metrics, using AWS Application Auto Scaling.

Policy types

Target tracking — keep a metric near a target, e.g. 60% average CPU.
Step scaling — change count by steps as a CloudWatch alarm breaches.
Scheduled — scale on a timetable for known traffic patterns.

Target tracking is the simplest and covers most needs.

Example

# Register the service as a scalable target, then track CPU at 60%
aws application-autoscaling register-scalable-target \
  --service-namespace ecs --resource-id service/prod/web \
  --scalable-dimension ecs:service:DesiredCount \
  --min-capacity 2 --max-capacity 10

aws application-autoscaling put-scaling-policy \
  --service-namespace ecs --resource-id service/prod/web \
  --scalable-dimension ecs:service:DesiredCount \
  --policy-name cpu60 --policy-type TargetTrackingScaling \
  --target-tracking-scaling-policy-configuration 'TargetValue=60.0,PredefinedMetricSpecification={PredefinedMetricType=ECSServiceAverageCPUUtilization}'

When to use it

A team registers their ECS service with Application Auto Scaling and sets a target-tracking policy on CPU utilization to add tasks automatically during peak hours.
An SRE adds a step scaling policy that adds 5 tasks when SQS queue depth exceeds 1000 messages so the processing service keeps pace with incoming work.
A developer sets minimum capacity to 2 and maximum to 50 on the ECS service so auto-scaling never goes below the HA minimum or above the cost ceiling.

More examples

Register ECS Service as Scalable Target

Registers the ECS service as an Application Auto Scaling target with a minimum of 2 and maximum of 20 tasks.

aws application-autoscaling register-scalable-target \
  --service-namespace ecs \
  --resource-id service/prod/api \
  --scalable-dimension ecs:service:DesiredCount \
  --min-capacity 2 \
  --max-capacity 20

Create CPU Target-Tracking Policy

Creates a target-tracking policy targeting 60% average CPU utilization with a 60-second scale-out cooldown and 5-minute scale-in cooldown.

aws application-autoscaling put-scaling-policy \
  --service-namespace ecs \
  --resource-id service/prod/api \
  --scalable-dimension ecs:service:DesiredCount \
  --policy-name cpu-tracking \
  --policy-type TargetTrackingScaling \
  --target-tracking-scaling-policy-configuration \
    'TargetValue=60.0,PredefinedMetricSpecification={PredefinedMetricType=ECSServiceAverageCPUUtilization},ScaleInCooldown=300,ScaleOutCooldown=60'

Create SQS-Based Step Scaling Policy

Adds 2 tasks when queue depth is 0-1000 messages above the threshold and 5 tasks when it exceeds 1000, with a 2-minute cooldown between scale-outs.

aws application-autoscaling put-scaling-policy \
  --service-namespace ecs \
  --resource-id service/prod/worker \
  --scalable-dimension ecs:service:DesiredCount \
  --policy-name sqs-step \
  --policy-type StepScaling \
  --step-scaling-policy-configuration \
    'AdjustmentType=ChangeInCapacity,StepAdjustments=[{MetricIntervalLowerBound=0,MetricIntervalUpperBound=1000,ScalingAdjustment=2},{MetricIntervalLowerBound=1000,ScalingAdjustment=5}],Cooldown=120'

Discussion

Be the first to comment on this lesson.