services.x-scaling ¶

Contents

services.x-scaling
- Range
- TargetScaling

This section allows to define scaling for the ECS Service. For SQS Based scaling using step scaling, refer to SQS Documentation.

         services:
  serviceA:
    x-scaling:
      Range: "1-10"
      TargetScaling:
        CpuTarget: 80

        

Range ¶

Range, defines the minimum and maximum number of containers you will have running in the cluster.

         #Syntax
# Range: "<min>-<max>"
# Example
Range: "1-21"

        

TargetScaling ¶

Allows you to define target scaling for the service based on CPU/RAM.

target scaling syntax reference ¶

          x-scaling:
  Range: "1-10"
  TargetScaling:
    CpuTarget: int (will be casted to float)
    MemoryTarget: int (will be casted to float)
    ScaleInCooldown: int (ie. 60)
    ScaleOutCooldown: int (ie. 60)
    DisableScaleIn: boolean (True/False)

         

Defines the CPU percentage that we want the service to be under. ECS will automatically create and adapt alarms to scale the service in/out so long as the average CPU usage remains beneath that value.

Attention

Note that setting both should not be set at the same time, as you might end up into a racing condition.

ScaleInCooldown / ScaleOutCooldown ¶

This allows you to define the Cooldown between scaling activities in order to limit drastic changes.

Hint

These are set only for the CPU and RAM targets, no impact on other scaling such as SQS.

DisableScaleIn ¶

Default: False

Same as the original Property in the CFN definition, this will deny a service to scale in after it has scaled-out for applications that do not support to scale-in.

Hint

If you define multiple services within the same family , the lowest value for CPU/RAM and highest for scale in/out are used in order to minimize the impact and focus on the weakest point.