Question

A DevOps team uses AWS CDK to define a multi-tier web application (VPC, ECS cluster, RDS database, and application services). They need to deploy this identical application to three separate AWS accounts (dev, staging, prod) within their organization. Each account must have its own isolated set of resources, and the team wants a single CDK codebase where the environment-specific account IDs and region are supplied at synth time, producing separate CloudFormation templates per environment. Which CDK approach best satisfies these requirements while keeping the code maintainable?

Accepted Answer

Define a reusable Stack subclass (or a Stage containing the stacks) and instantiate it once per environment in the CDK app, passing each environment's account and region via the stack/stage 'env' property.

Answer

Create a single CloudFormation StackSet from the CDK synthesized template and let StackSets fan the deployment out to all three accounts automatically.

Answer

Use CDK context lookups with cdk.context.json to hardcode all three account IDs into one stack, then deploy that one stack three times with --force.

Answer

Define one stack with no explicit 'env' property so it becomes environment-agnostic, then rely on the CLI credentials of each account to place resources correctly at deploy time.