Architectural drift is a form of architectural regression often caused by inappropriate evolution that leads to the loss of characteristics that were once met.

As architectures grow in size and complexity, it becomes more and more difficult to predict the effects of a change, therefore drift isn’t necessarily due to incompetence or lack of skill.

Much hard work, time and money are invested to meet architectural aims and measures can be taken to ensure this does not go in vain with the passing of time and system growth.

𝐅𝐢𝐭𝐧𝐞𝐬𝐬 𝐅𝐮𝐧𝐜𝐭𝐢𝐨𝐧𝐬: Fitness functions defend against architectural drift by providing change resilience in an evolutionary-architecture compatible way, ensuring that changes are predictable.

Fitness functions are defined as: “𝘈𝘯𝘺 𝘮𝘦𝘤𝘩𝘢𝘯𝘪𝘴𝘮 𝘵𝘩𝘢𝘵 𝘱𝘳𝘰𝘷𝘪𝘥𝘦𝘴 𝘢𝘯 𝘰𝘣𝘫𝘦𝘤𝘵𝘪𝘷𝘦 𝘪𝘯𝘵𝘦𝘨𝘳𝘪𝘵𝘺 𝘢𝘴𝘴𝘦𝘴𝘴𝘮𝘦𝘯𝘵 𝘰𝘧 𝘴𝘰𝘮𝘦 𝘢𝘳𝘤𝘩𝘪𝘵𝘦𝘤𝘵𝘶𝘳𝘦 𝘤𝘩𝘢𝘳𝘢𝘤𝘵𝘦𝘳𝘪𝘴𝘵𝘪𝘤 𝘰𝘳 𝘤𝘰𝘮𝘣𝘪𝘯𝘢𝘵𝘪𝘰𝘯 𝘰𝘧 𝘢𝘳𝘤𝘩𝘪𝘵𝘦𝘤𝘵𝘶𝘳𝘦 𝘤𝘩𝘢𝘳𝘢𝘤𝘵𝘦𝘳𝘪𝘴𝘵𝘪𝘤𝘴” – Building Evolutionary Architectures

Evolutionary Architecture is defined as: “𝘈𝘯 𝘦𝘷𝘰𝘭𝘶𝘵𝘪𝘰𝘯𝘢𝘳𝘺 𝘢𝘳𝘤𝘩𝘪𝘵𝘦𝘤𝘵𝘶𝘳𝘦 𝘴𝘶𝘱𝘱𝘰𝘳𝘵𝘴 𝘨𝘶𝘪𝘥𝘦𝘥, 𝘪𝘯𝘤𝘳𝘦𝘮𝘦𝘯𝘵𝘢𝘭 𝘤𝘩𝘢𝘯𝘨𝘦 𝘢𝘤𝘳𝘰𝘴𝘴 𝘮𝘶𝘭𝘵𝘪𝘱𝘭𝘦 𝘥𝘪𝘮𝘦𝘯𝘴𝘪𝘰𝘯𝘴”. – Building Evolutionary Architectures

Fitness functions not only provide a form of protection against regression but they can also be used to adopt fitness function-driven development.

𝐅𝐢𝐭𝐧𝐞𝐬𝐬 𝐅𝐮𝐧𝐜𝐭𝐢𝐨𝐧-𝐃𝐫𝐢𝐯𝐞𝐧 𝐃𝐞𝐯𝐞𝐥𝐨𝐩𝐦𝐞𝐧𝐭: FFDD Is the equivalent of TDD for non-functional requirements such as scalability and other architectural characteristics, aka “-𝘪𝘭𝘪𝘵𝘪𝘦𝘴”. In other words, this approach allows you to define an architectural aim and then work your way towards meeting it.

Governance can also be achieved via the application of fitness function tests by integrating them into CI pipelines allowing for early detection of compromising changes.

Any testing framework can be utilised to perform fitness function testing and could be a collection of frameworks or tools that work in conjunction with each other, for example, code analysis tools can be used to address security concerns while load testing tools can be used to address performance constraints.

Careful consideration should be made when adopting fitness functions, such as the impact on CI build times and operational expenditure. If such concerns exist then it may be best to run such tests at lower frequencies but bear in mind that ease of rectification correlates with early detection.

Image Credits: Thoughtworks

The below image illustrates FF test results, these test results can be used to drive release and code review approvals.

The below shows an FF test example targetted towards performance