Apart from being able to create normal “24/7” type instances in your AWS Opsworks Stack, you can also configure a Load or Time based autoscaling instances to suit your needs.

A LoadBased instance will come to life when the average CPU/Memory % of your normal instances go beyond a certain threshold. The load to your application (typically from ELB) will then be shared by 1 or more of these autoscaling instances.

Here are the steps required to configure it manually:

  • In your chosen Opsworks stack, in the “Instances” section, choose “Load-based” subsection. There, ‘Edit Configuration’ to define your threshold.
  • Click the “+ Instance” or Add Instance button in the “Load-based” subsection page. Define your Instance settings, and click “Add Instance”.

This instance will not boot and come to online right away. Only when the average load among the normal 24/7 servers go up, this instance will boot and come online to save the day.

Cloudformation Setup

Here’s what you have to do to add autoscaling instance in your cloudformation template.

1) Define the “LoadbasedAutoScaling” property in your “AWS::OpsWorks::Layer” like so:

"LoadBasedAutoScaling": {
  "Enable": "true",
  "DownScaling": {
    "InstanceCount": 1,
    "ThresholdsWaitTime": 5,
    "IgnoreMetricsTime": 10,
    "CpuThreshold": 10.0,
    "MemoryThreshold": 15.0
  "UpScaling": {
    "InstanceCount": 1,
    "ThresholdsWaitTime": 5,
    "IgnoreMetricsTime": 15,
    "CpuThreshold": 80.0,
    "MemoryThreshold": 85.0

2) While defining the Opsworks Instance, add the “AutoScalingType” property, and set it to “load”:

"OpsInstance": {
  "Type": "AWS::OpsWorks::Instance",
  "Properties": {
    "Architecture": "x86_64",
    "InstallUpdatesOnBoot": "true",
    "InstanceType": "t2.micro",
    "Os": "Ubuntu 14.04 LTS",
    "RootDeviceType": "ebs",
    "SshKeyName": "somekey",
    "StackId": {
      "Ref": "OpsStack"
    "SubnetId": {
      "Ref": "Subnet1"
    "LayerIds": [
        "Ref": "OpsLayer"
    "AutoScalingType": "load"