Set Up Autoscaling

You can enable the autoscaling feature in SaladCloud through the portal or via the API. Here’s how:

​

Using the Portal

1. Create or Edit a Container Group:

   • In the SaladCloud portal, create a new container group or edit an existing one.
   • Note: You can only configure your Job Queue connection when creating a new container group. If the container is not connected to a job queue during creation, you will not be able to retrospectively add the job queue connection later through the “Edit Container” wizard. This differs from the autoscaling settings, which can be added or edited later for any container group that has an active queue connection.

2. Enable the Job Queue:

   • Ensure that your container group is configured to use the Job Queue, which the autoscaler monitors to determine scaling behavior. You can find detailed instructions on how to enable the Job Queue here

3. Configure Autoscaling Settings:

• Under the Job Queue section, select Enable Autoscaling to reveal the autoscaling settings.

• Under the Autoscaling section, configure the desired settings such as Desired Queue Length, Minimum Replicas, Maximum Replicas, Period, and optional settings like Maximum Upscale Per Minute and Maximum Downscale Per Minute. Check the Autoscaling Settings for more details.

4. Save and Deploy:
   • After configuring the settings, save the changes and deploy the container group. The autoscaler will begin monitoring the queue and scaling container instances after the first message is sent to the queue.

Note: You still need to specify replica count in the container group configuration. This number will be used the first message is sent to the queue. Once that happens autoscaller will take over and adjust the replica count based on queue length.

​

Using the API

You can also configure autoscaling programmatically using SaladCloud’s API.

1. Create a Container Group: Use the SaladCloud API to create a container group. Ensure that the Job Queue is enabled in the configuration.

2. Set Autoscaling Parameters: Set the following parameters in the API request body to enable autoscaling. Example:

   "queue_autoscaler": {
      "min_replicas": 3,
      "max_replicas": 20,
      "desired_queue_length": 5,
      "polling_period": 15,
      "max_upscale_per_minute": 20,
      "max_downscale_per_minute": 10
    }
   

   For the full api call example, refer to the API documentation.