Integration testing

Unit tests focus on the smallest parts of the system. However, testing components in isolation sometimes is not enough. This is especially true when the code under test goes beyond the system’s borders and uses other (often external) components. Integration testing is the test level we use to test the integration between our code and external parties.

Let’s consider a real-world example. Software systems commonly rely on database systems. To communicate with the database, developers often create a class whose only responsibility is to interact with this external component (think of Data Access Object [DAO] classes). These DAOs may contain complicated SQL code. Thus, a tester feels the need to test the SQL queries. The tester does not want to test the entire system, only the integration between the DAO class and the database. The tester also does not want to test the DAO class in complete isolation. After all, the best way to know whether a SQL query works is to submit it to the database and see what the database returns.

This is an example of an integration test. Integration testing aims to test multiple components of a system together, focusing on the interactions between them instead of testing the system as a whole (see figure 1.6). Are they communicating correctly? What happens if component A sends message X to component B? Do they still present correct behavior?

Figure 1.6 Integration testing. Our goal is to test whether our component integrates well with an external component.

Integration testing focuses on two parts: our component and the external component. Writing such a test is less complicated than writing a test that goes through the entire system and includes components we do not care about.

Compared to unit testing, integration tests are more difficult to write. In the example, setting up a database for the test requires effort. Tests that involve databases generally need to use an isolated instance of the database just for testing purposes, update the database schema, put the database into a state expected by the test by adding or removing rows, and clean everything afterward. The same effort is involved in other types of integration tests: web services, file reads and writes.