Learning new ways to test your infrastructure

Once feature flags have been implemented on the backend systems, in addition to the frontend applications, some new testing opportunities can make use of controlling what downstream systems are being requested. By this, I mean that a client application's use of LaunchDarkly can target a different endpoint or configuration so that a different backend service is called for testing. This is not the type of testing that will validate that components function as expected or that systems integrate well; instead, this testing relies on the fact that the production environment itself can be safely tested. In previous chapters, especially Chapter 5, Experimentation, the focus was on being able to evaluate the features of the production environment. With this approach to testing, it is also possible to assess the infrastructure itself.

When discussing performing migrations in the previous section, we explained that different endpoints could...