Anjali Panwar

2 minute read


We’re excited to announce our next application which supports Amazon’s ECS (EC2 Container Services). CMPUTE has launched the direct support for running your ECS cluster. Now, you can run your ECS Cluster hassle free using our platform with spot instances/fleet and zero down time.

You can use our ECS app to schedule the placement of container instances across your cluster. Cmpute’s ECS app eliminates the need for you to operate container instance management. Once you create an ECS and imports that in cmpute, it is cmpute’s responsibility to launch and manage all the container instance across the clusters.

Cmpute also gives flexibility to:

  • Schedule the execution of ECS clusters.
  • Scaling the container instances behind Auto Scaling Group.

The following guide gives the step by step approach of managing your ECS cluster through the CMPUTE platform.

During ECS cluster creation, AWS creates an auto scaling group for the cluster. If you are running a ECS cluster with Name “ProdApp”, there will be a corresponding AWS autoscaling group in the format


For more detail: Click here.

Step 1: Create a CloudWatch Alarm for the Metric For this setup, you create an alarm on the cluster MemoryReservation metric, which will alert you when the cluster’s memory reaches above certain threshold.

Step 2: Login to your account and select ECS App.

Step 3: Select your AWS account and the Region and fill the following details:

  • Job Name
  • Account
  • Cloud Region
  • Auto Scaling Group name
  • Min OnDemand instances


Step 4: Now either schedule or run the job.


CMPUTE will launch the specified number of instances and associate with the ECS cluster. Continuously monitor and manage entire cluster. If any of the scaling thresholds are hit, CMPUTE will scale up/ down instances.

Voila ! now you have successfuly configured ECS cluster with CMPUTE !!!!

Note: Currently, cmpute supports ECS which is created with ‘On-Demand’ provisioning model. Soon, we will be launching ECS support with ‘Spot Instance’ provisioning model.

comments powered by Disqus