Customer-managed environments

With customer-managed environments, you can use existing infrastructure, like a VPC, that you already have deployed as your AWS Proton environment. While using customer-managed environments, you can provision your own shared resources outside of AWS Proton. However, you can still allow AWS Proton to consume relevant provisioning outputs as inputs for AWS Proton services when they are deployed. If the outputs can change, AWS Proton is able to accept updates. AWS Proton is unable to change the environment directly, though, since the provisioning is managed outside of AWS Proton.

After the environment is created, you're responsible for providing the same outputs to AWS Proton that would have been created if AWS Proton had made the environment, such as Amazon ECS cluster names or Amazon VPC IDs.

With this functionality, you can deploy and update AWS Proton service resources from an AWS Proton service template to this environment. However, the environment itself isn't modified through template updates in AWS Proton. You're responsible for executing updates to the environment and updating those outputs in AWS Proton.

You can have multiple environments in a single account that are a mix of AWS Proton managed and customer-managed environments. You can also link a second account and use an AWS Proton template in the primary account to execute deployments and updates to environments and services in that second, linked account.

How to use customer-managed environments

The first thing administrators need to do is register an imported, customer-managed environment template. Don't supply manifests or infrastructure files in the template bundle. Only supply the schema.

The schema below outlines a list of outputs using the open API format and replicates the outputs from an AWS CloudFormation template.

The following example is a snippet of the output sections of an AWS CloudFormation template for a corresponding Fargate template.

Outputs:
  ClusterName:
    Description: The name of the ECS cluster
    Value: !Ref 'ECSCluster'
  ECSTaskExecutionRole:
    Description: The ARN of the ECS role
    Value: !GetAtt 'ECSTaskExecutionRole.Arn'
  VpcId:
    Description: The ID of the VPC that this stack is deployed in
    Value: !Ref 'VPC'
[...]

The schema for the corresponding AWS Proton imported environment is similar to the following. Don't supply default values in the schema.

schema:
  format:
    openapi: "3.0.0"
  environment_input_type: "EnvironmentOutput"
  types:
    EnvironmentOutput:
      type: object
      description: "Outputs of the environment"
      properties:
        ClusterName:
          type: string
          description: "The name of the ECS cluster"
        ECSTaskExecutionRole:
          type: string
          description: "The ARN of the ECS role"
        VpcId:
          type: string
          description: "The ID of the VPC that this stack is deployed in"
[...]

At the time of registering the template, you indicate that this template is imported and provides the Amazon S3 bucket location for the bundle. AWS Proton validates that the schema only contains environment_input_type and no AWS CloudFormation template parameters before putting the template in draft.

You can provide both of these through either the console or the AWS CLI using a process similar to the deployment of a regular environment.