View a markdown version of this page

Tutorial: Create a managed compute environment using Amazon ECS Managed Instances - AWS Batch
PrerequisitesCreating an Amazon ECS Managed Instances compute environment (console)Creating an Amazon ECS Managed Instances compute environment (AWS CLI)Next steps

Tutorial: Create a managed compute environment using Amazon ECS Managed Instances

This tutorial walks you through creating an AWS Batch compute environment that uses Amazon ECS Managed Instances. Amazon ECS manages the Amazon EC2 instances on your behalf — you specify the maximum capacity and optional instance constraints, and Amazon ECS handles provisioning, scaling, and termination.

Prerequisites

Before you create an Amazon ECS Managed Instances compute environment, you must have the following:

  • An Amazon ECS infrastructure role — an IAM role that Amazon ECS assumes to manage Amazon EC2 instances on your behalf. The role must have a trust policy for ecs.amazonaws.com. For more information, see Amazon ECS infrastructure IAM role in the Amazon Elastic Container Service Developer Guide.

  • An Amazon EC2 instance profile — an instance profile that uses the AmazonECSInstanceRolePolicyForManagedInstances managed policy. The trust policy must allow ec2.amazonaws.com to assume the role.

  • Your IAM principal must have iam:PassRole permission for the infrastructure role with the condition iam:PassedToService: ecs.amazonaws.com.

  • A VPC with subnets that have outbound internet access (either public IP assignment or a NAT gateway).

Creating an Amazon ECS Managed Instances compute environment (console)

  1. Open the AWS Batch console at https://console.aws.amazon.com/batch/.

  2. From the navigation bar, choose the AWS Region to use.

  3. In the left navigation pane, choose Environments.

  4. Choose Create environment, then choose Create compute environment.

  5. For Compute environment type, choose ECS Managed Instances.

  6. For Compute environment name, enter a unique name. The name can be up to 128 characters long and can contain uppercase and lowercase letters, numbers, hyphens (-), and underscores (_).

  7. For Maximum vCPUs, enter the maximum number of vCPUs that the compute environment can scale to.

  8. For Infrastructure role, choose the Amazon ECS infrastructure role that you created as a prerequisite.

  9. For Instance profile, choose the Amazon EC2 instance profile that you created as a prerequisite.

  10. In the Networking section:

    1. For VPC, choose the VPC where instances will be launched.

    2. For Subnets, choose one or more subnets. The subnets must have outbound internet access.

    3. For Security groups, choose one or more security groups to associate with the instances.

  11. (Optional) For Capacity type, choose Spot to use Spot Instances for cost savings on fault-tolerant workloads. The default is On-Demand.

  12. (Optional) For Allowed instance types, specify which instance types Amazon ECS can launch. If you don't specify any, all available instance types are eligible.

  13. Choose Create compute environment.

Creating an Amazon ECS Managed Instances compute environment (AWS CLI)

Use the create-compute-environment command to create an Amazon ECS Managed Instances compute environment.

Basic example

The following example creates a compute environment with the minimum required configuration. Amazon ECS can use any available instance type.

$ aws batch create-compute-environment \
    --compute-environment-name my-managed-instances-ce \
    --type MANAGED \
    --state ENABLED \
    --compute-resources '{
      "type": "ECS_MANAGED_INSTANCES",
      "maxvCpus": 256,
      "managedInstancesProvider": {
        "infrastructureRoleArn": "arn:aws:iam::123456789012:role/ecsInfrastructureRole",
        "instanceLaunchTemplate": {
          "ec2InstanceProfileArn": "arn:aws:iam::123456789012:instance-profile/ecsInstanceProfile",
          "networkConfiguration": {
            "subnets": ["subnet-abcde012", "subnet-bcde012a"],
            "securityGroups": ["sg-abcde012"]
          }
        }
      }
    }'

Spot with specific instance types

The following example creates a Spot-backed compute environment constrained to specific instance types.

$ aws batch create-compute-environment \
    --compute-environment-name my-spot-managed-instances-ce \
    --type MANAGED \
    --state ENABLED \
    --compute-resources '{
      "type": "ECS_MANAGED_INSTANCES",
      "maxvCpus": 1000,
      "managedInstancesProvider": {
        "infrastructureRoleArn": "arn:aws:iam::123456789012:role/ecsInfrastructureRole",
        "instanceLaunchTemplate": {
          "ec2InstanceProfileArn": "arn:aws:iam::123456789012:instance-profile/ecsInstanceProfile",
          "networkConfiguration": {
            "subnets": ["subnet-abcde012", "subnet-bcde012a"],
            "securityGroups": ["sg-abcde012"]
          },
          "instanceRequirements": {
            "allowedInstanceTypes": ["m5.large", "m5.xlarge", "m6i.large", "m6i.xlarge"]
          },
          "capacityOptionType": "SPOT"
        }
      }
    }'

Output:

{
    "computeEnvironmentName": "my-spot-managed-instances-ce",
    "computeEnvironmentArn": "arn:aws:batch:us-east-1:123456789012:compute-environment/my-spot-managed-instances-ce"
}

With capacity reservations

The following example creates a compute environment that targets On-Demand Capacity Reservations for predictable capacity.

$ aws batch create-compute-environment \
    --compute-environment-name my-reserved-managed-instances-ce \
    --type MANAGED \
    --state ENABLED \
    --compute-resources '{
      "type": "ECS_MANAGED_INSTANCES",
      "maxvCpus": 512,
      "managedInstancesProvider": {
        "infrastructureRoleArn": "arn:aws:iam::123456789012:role/ecsInfrastructureRole",
        "instanceLaunchTemplate": {
          "ec2InstanceProfileArn": "arn:aws:iam::123456789012:instance-profile/ecsInstanceProfile",
          "networkConfiguration": {
            "subnets": ["subnet-abcde012", "subnet-bcde012a"],
            "securityGroups": ["sg-abcde012"]
          },
          "instanceRequirements": {
            "allowedInstanceTypes": ["m5.xlarge", "m5.2xlarge"]
          },
          "capacityReservations": {
            "reservationGroupArn": "arn:aws:ec2:us-east-1:123456789012:capacity-reservation-group/my-reservation-group",
            "reservationPreference": "RESERVATIONS_FIRST"
          }
        }
      }
    }'

Next steps

After creating your compute environment, complete the following steps to run jobs:

  1. Create a job queue and associate your Amazon ECS Managed Instances compute environment with it. For more information, see Create a job queue.

  2. Create a job definition with platformCapabilities set to MANAGED_INSTANCES and using ecsProperties. For more information, see Job definitions on Amazon ECS Managed Instances.

  3. Submit a job to the job queue. For more information, see Tutorial: submit a job.