Create Predictive Scaling Policy

STOP Please note: This workshop version is now deprecated, and an updated version has been moved to AWS Workshop Studio. This workshop remains here for reference to those who have used this workshop before for reference only. Link to updated workshop is here: Efficient and Resilient Workloads with Amazon EC2 Auto Scaling.

Following up with your scenario, one of the requirements is to reduce the impact of time the application takes to become ready.

Predictive scaling has a SchedulingBufferTime parameter that allows instances to launch in advance. For example, the forecast says to add capacity at 10:00 AM, and you choose to pre-launch instances by 5 minutes. In that case, the instances will be launched at 9:55 AM. The intention is to give the instances time to be initialized..

Create the predictive scaling policy

  1. In Cloud9 IDE terminal, check you’re at this directory ec2-spot-workshops/workshops/efficient-and-resilient-ec2-auto-scaling

  2. Review the policy configuration file and note how the custom metrics have been used in it.

cat ./policy-config.json

Note this section in the outcome, what effect do you think the parameter MaxCapacityBreachBehavior could have on the Auto Scaling group capacity?

Show answer

    "Mode": "ForecastAndScale",
    "SchedulingBufferTime": 300,
    "MaxCapacityBreachBehavior": "HonorMaxCapacity"
  1. Run this command to create the policy with the custom metrics and attach it to the auto scaling group.
aws autoscaling put-scaling-policy --policy-name workshop-predictive-scaling-policy \
  --auto-scaling-group-name "ec2-workshop-asg" --policy-type PredictiveScaling \
  --predictive-scaling-configuration file://policy-config.json

If successful, the command should return the created policy ARN.

    "PolicyARN": "arn:aws:autoscaling:ap-southeast-2:115751184547:scalingPolicy:df0e550e-b0d6-4924-8663-d394de77b0e3:autoScalingGroupName/ec2-workshop-asg:policyName/workshop-predictive-scaling-policy",
    "Alarms": []

To edit a predictive scaling policy that uses customized metrics, you must use the AWS CLI or an SDK. Console support for customized metrics will be available soon.