System tests which test an application that is deployed to a production-like environment in an end-to-end fashion provide the most representative verification. Since they run in a later step in the Continuous Delivery pipeline, they provide slower feedback. Test cases, such as verifying the JSON mappings of an HTTP endpoint, will take longer before they provide feedback to engineers.

Tackling and maintaining complex test scenarios is an aspect that takes quite some time and effort. Enterprise applications require definition and maintenance of test data and configuration, especially when many external systems are involved. For example, an end-to-end test that verifies creating a car in the car manufacture application requires external concerns such as the assembly line to be set up as well as test data. Managing these scenarios involves a certain effort.

End-to-end tests attempt to use external systems and databases similarly to production. This introduces the challenge to handle unavailable or erroneous environments. External systems or databases that are unavailable cause the tests to fail; however, the application is not responsible for this test failure. This scenario violates the requirement of predictability, that tests should not depend on external factors that provide false positives. Therefore, it's advisable that system tests mock away external systems that are not part of the application under test. Doing this enables to construct predictable end-to-end tests. The Sub-chapter System tests covers how to implement this approach.