When developing for autoscale, it's important to keep a few key autoscale concepts in mind. For example, it's important to note that a resource can have only one autoscale setting. That said, an autoscale setting can have multiple profiles, which can each in turn have multiple autoscale rules defined on. Autoscale settings, incidentally, scale instances horizontally or out by increasing the number of instances. Instances are scaled in by decreasing the number of instances. Autoscale settings are configured with a maximum number of instances, a minimum number of instances, and a default number of instances.

 All autoscale thresholds are calculated at the instance level. For example, a setting of scale out by one instance when average processor utilization is greater than 80% when instance count is two means to scale out when the average CPU utilization across all instances is over 80%. Remember that all autoscale failures are logged to the Activity Log. As such, you can configure Activity Log alerts that will notify you via email, SMS, or even webhooks whenever there is an autoscale failure. Quite similarly, all successful scale actions are posted to the Activity Log as well. Configuring an Activity Log alert allows you to be notified via email, SMS, or webhooks whenever there is a successful autoscale action. By keeping these types of autoscale concepts in mind when developing, you can ensure a smoother development process and fewer headaches down the road.