Workload dependencies across environments - Organizing Your AWS Environment Using Multiple Accounts

Workload dependencies across environments

When you consider the structure of your workload-oriented OUs, you should decide on the extent to which you expect access between production and non-production environments.

Production environments accessing non-production

Generally, workloads deployed to your production environments should not depend on workloads contained in your non-production environments.

Non-production environments accessing dependencies

In non-production environments, it is common for workloads to depend on stable shared application, data, and infrastructure services. Where feasible, we recommend that these shared services be non-production test instances. These non-production test instances should use test data so that your non-production workloads do not depend on access to your production environments and data.

For example, you can configure workloads in a non-production test environment that depend on integrating with a data service to use a stable, shared test instance of the service that is populated with test data.

However, in some cases non-production environments might need access to production shared services. For example, it’s typical for non-production development and test environments to require read-only access to shared source code and artifact management services. Providing access to these shared services enables you to deploy candidate and promoted changes and artifacts to your non-production environments in support of development and testing activities.