Skip to main content

Autoscaling

Autoscaling means automatically increasing or decreasing the number of instances based on defined rules. For example, you can add 2 instances when an API receives more than 1,000 requests within an hour.

When autoscaling is enabled, the Autoscaling tab appears on the endpoint management screen.

Autoscaling conditions

Adding a Scale Condition​

Click Add Scale Condition to create a new autoscaling rule.

Scale Condition​

The available fields are described below.

  • Metric
    • The metric used to decide whether to scale instances up or down.
    • You can pick total requests (total_requests), requests per second (requests_per_sec), request latency (latency_ms), and similar metrics.
  • Target value and operator
    • Compare the metric value against a target using an operator.
    • For example, for the total_requests metric you can choose "greater than or equal to" with a target value of "50".
  • Measurement window
    • The period over which the metric value is collected.
    • For example, with total_requests and a 10-minute window, the rule counts all requests in the last 10 minutes.
  • Cooldown
    • The amount of time after a scaling action during which further scaling is suppressed.
    • Use this to keep autoscaling from triggering too frequently.
  • Action
    • The action to perform — for example, scaling instances up or down.
    • For example, setting "decrease instances" by "1" decreases the instance count by 1 whenever the metric condition is met.

Autoscaling History​

A record of times the autoscaling conditions you defined were triggered. Use this to see when each condition fired.