I’ve written about how overloaded the term “load test” is, and how we might spin off some of its meanings into other useful terms. The first was the endurance test.
The second spin-off can be the recovery test.
The idea of the recovery test is to deprive the application of a critical resource, then determine whether its behavior is acceptable.
Some such critical resources:
- Disk space
- Network connection
- Database connections
- Database integrity
- Data is retained, especially user data.
- The problem is logged as fully as possible.
- Users are informed of the problem in a timely manner.
- The application restarts, if necessary and appropriate.
- The application resumes previous activities, if necessary and appropriate.