Use the App AutoScaler

In this topic you will learn how to use the App AutoScaler. Cloud Foundry offers you simple elasticity of your applications. You can easily scale the application instances up and down as you like. The App AutoScaler service allows to even automate this process of scaling application instances up and down based on a ruleset you can create on your own.

The App AutoScaler service component is based on an open source project. You can check out its code in the public App AutoScaler repository.

Do you just want to try out the STACKIT Cloud Foundry for a first time, without having your actual cloud ready application prepared yet?

Then you can just use one of the sample applications the Cloud Foundry Open Source Community created for demo use cases, which you can find in their GitHub repository.

App AutoScaler dashboard in Stratos

The App AutoScaler provides the capability to adjust the computation resources for Cloud Foundry applications through:

• Dynamic scaling based on application performance metrics
• Scheduled scaling based on time

App AutoScaler ensures your application to have the right number of instances to maintain the service quality and pay the resource you really need. With auto-scaling capability, your application gains the following benefits:

• Better service quality: Auto-scaling monitors your application performance and scales out/in your application horizontally to match the resource need. In this way, your service quality target defined in the policy can be maintained.
• Better availability: Auto-scaling helps prepare enough number of application instances in advance if there is predictable workload spike in a given period of time. It will avoid overloading and application crash during the spike.
• Cost reduction: Auto-scaling allows you to pay for the resource you really need. You don’t need to have capacity planning and deploy your application with a fixed number of instances. It saves you money by starting new instances when your application needs them and destroys them when it does not.

App AutoScaler is offered as a Cloud Foundry service via Open Service Broker API, you need to provision and bind App AutoScaler service through Cloud Foundry CLI or the Portal first.

Create an instance of the service

cf create-service autoscaler autoscaler-free-plan <service-instance_name>

Bind the service instance to your application

You can attach scaling policy together with service binding by providing the policy file name as a parameter of the service binding command.

cf bind-service <app_name> <service_instance_name> -c <policy_file_name>

To disconnect App AutoScaler from your application, unbind the service instance. This will remove the autoscaling policy as well. Furthermore, you can deprovision the service instance if no application is bound.

cf unbind-service <app_name> <service_instance_name>

cf delete-service <service_instance_name>

App AutoScaler policy settings

Sample scaling policy JSON-file:

{
  "instance_min_count": 1,
  "instance_max_count": 4,
  "scaling_rules": [
    {
      "metric_type": "cpu",
      "threshold": 5,
      "operator": ">=",
      "adjustment": "+1"
    },
    {
     "metric_type": "cpu",
     "threshold": 5,
     "operator": "<",
     "adjustment": "-1"
    }
  ]
}

App AutoScaler schemas

App AutoScaler requires a policy file written in JSON with the following schemas:

Policy

Expand

Name	Type	Required	Description
instance_min_count	int	true	minimum number of instance count
instance_max_count	int	true	maximum number of instance count
scaling_rules	JSON-array<scaling_rules>		dynamic scaling rules
schedules	JSON-array		scheduled

Scaling rules

Expand

Name	Type	Required	Description
metric_type	string	true	See list below.
threshold	int	true	The boundary when metric value exceeds is considered as a breach.
operator	string	true	>, <, >=, <=
adjustment	string	true	The adjustment approach for instance count with each scaling. Support regex format ^[-+][1-9]+[0-9]*[%]?$, i.e. +5 means adding 5 instances, -50% means shrinking to the half of current size.
breach_duration_secs	int, seconds	false	Time duration to fire scaling event if it keeps breaching.
cool_down_secs	int, seconds	false	The time duration to wait before the next scaling kicks in.

Metric_type values

• cpu - short name of “CPU utilization”, is the CPU usage of your application in percentage.
• memoryused - represents the absolute value of the used memory of your application. The unit of memoryused metric is “MB”.
• memoryutil - short name of “memory utilization”, is the used memory of the total memory allocated to the application in percentage. For example, if the memory usage of the application is 100MB and memory quota is 200MB, the value of memoryutil is 50%.
• responsetime - represents the average amount of time the application takes to respond to a request in a given time period. The unit of responsetime is “ms” (milliseconds).
• throughput - is the total number of processed requests in a given time period. The unit of throughput is “rps” (requests per second).
• Custom metric - You can define your own metric name and emit your own metric to App AutoScaler to trigger further dynamic scaling. Only alphabet letters, numbers and ”_” are allowed for a valid metric name, and the maximum length of the metric name is limited up to 100 characters.

Scheduled

Expand

Name	Type	Required	Description
timezone	string	true	using timezone definition of Java
recurring_schedule	JSON-array<recurring_schedules>		the schedules which will take effect repeatedly, see Recurring Schedule below
specific_date	JSON-array<specific_date>		the schedules which take effect only once, see Specific Date below

Recurring schedule

Expand

Name	Type	Required	Description
start_date	string, “yyyy-mm-dd”	false	the start date of the schedule; must be a future time
end_date	string, “yyyy-mm-dd”	false	the end date of the schedule; must be a future time.
start_time	string, “hh:mm”	true	the start time of the schedule
end_time	string, “hh:mm”	true	the end time of the schedule
days_of_week / days_of_month	array	false	recurring days of a week or month; use [1,2,..7] or [1,2,..31] to define it
instance_min_count	int	true	minimum number of instance count for this schedule
instance_max_count	int	true	maximum number of instance count for this schedule
initial_min_instance_count	int	false	the initial minimal number of instance count for this schedule

If one schedule overlaps another, the one which starts first will be guaranteed, while the later one is completely ignored, for example:

Schedule #1: ———sssssssssss——————————————
Schedule #2: ——————————ssssssssss————————
Schedule #3: ————————————————sssssssss———

With the above definition, schedule #1 and #3 will be applied, while schedule #2 is ignored.