[20180425220726] Service Autoscaling and Bug Fixes

This release adds autoscaling at the application level. It can be defined like this:

      count: 1-10
        cpu: 70
        memory: 90
        requests: 200

The following targets are available:

  • cpu: Average CPU utilization (%) of the processes of this service
  • memory: Average CPU utilization (%) for the processes of this service
  • requests: Requests per minute per process of this service
  • custom: Use any Cloudwatch metric (see the docs for more details)

The service autoscaler will determine the maximum number of individual processes needed to achieve the defined targets and automatically adjust scale within the count: range to meet them.