How can a system pass its acceptance tests, and then still not turn out to be satisfactory?
No matter how much testing is performed, if the test coverage is not adequate the system may contain unrevealed defects or insufficiencies.
Thus the buyers of a system must have input into the acceptance test design and content. This should happen at two points:
- The initial requirements and specifications.
- The review of the test plan.
The acceptance test must demonstrate that the system will perform adequately in the way that the system is intended to be used. Before the system goes live, final tests must be done with the system configured for live use, using real-world data and realistic test scenarios.
There are many types of tests the should be performed well before the “final” acceptance test.
Early testing includes proof-of-concept tests, stress and load tests, performance tests, failure tests, and functional tests.
Fault tolerance, hot or cold standby, and backup & restore capabilities must be tested in realistic scenarios.
Network performance testing is required as well, both before the system goes live (in a test environment) and after it is in live use.
Individual component testing (also called “unit testing”) tests various components individually, integration testing tests that separate systems work together properly, end-to-end testing that tests the overall system in a situation that mimics real-world use.
All these tests should be done BEFORE the system “goes live”.
There is also operational testing, which is a period of time (usually 30 to 90 days) in which the live system is exercised in normal operations and real or simulated incident response and investigations.
- When and where do you do how much of each?
- How much testing is too much? Or too little?
The buyers and users of the system will want to focus on how they plan to use the system on a daily basis. This is good, but it’s usually not enough.
Someone expert in the types of tests required must see that the test plan contains:
- a valid test approach and schedule
- appropriate test cases
- reliable test procedures
- correct test system configuration
- proper test environment
- clear description of expected test outcomes
- valid test pass/fail criteria
- tests appropriate for the technologies employed
- proper testing of packaged software components
The small investment of time that will be involved in reviewing (and if necessary, revising) the test plan will make a big contribution towards a successful system turnover.
RBCS system testing specialists can review your project’s test plan, or write one for your project. Call us at 949-831-6788.