As software becomes more critical to modern business, software needs to be able to react quickly to changes, allowing new features to be conceived, developed and put into production rapidly. The techniques of agile software development began in the 1990s and became steadily more popular in the last decade. They focus on a flexible approach to planning, which allows software products to change direction as the users’ needs change and as product managers learn more about how to make their users effective. While widely accepted now, agile approaches are not easy, requiring significant skills for a team, but more importantly a culture of open collaboration both within the team and with a team’s partners.

This need to respond fluently to changes has an important impact upon the architecture of a software system. The software needs to be built in such a way that it is able to adapt to unexpected changes in features. One of the most important ways to do this is to write clear code, making it easy to understand what the program is supposed to do. This code should be divided into modules which allow developers to understand only the parts of the system they need to make a change. This production code should be supported with automated tests that can detect any errors made when making a change while providing examples of how internal structures are used. Large and complex software efforts may find the microservices architectural style helps teams deploy software with less entangling dependencies.

Creating software that has a good architecture isn’t something that can be done first time. Like good prose, it needs regular revisions as programmers learn more about what the product needs to do and how best to design the product to achieve its goals. Refactoring is an essential technique to allow a program to be changed safely. It consists of making small changes that don’t alter the observable behavior of the software. By combining lots of small changes, developers can revise the software’s structure supporting significant modifications that weren’t planned when the system was first conceived.

Software that runs only on a developer’s machine isn’t providing value to the customers of the software. Traditionally releasing software has been a long and complicated process, one that hinders the need to evolve software quickly. Continuous delivery uses automation and collaborative workflows to remove this bottleneck, allowing teams to release software as often as the customers demand. For Continuous delivery to be possible, we need to build in a solid foundation of Testing, with a range of automated tests that can give us confidence that our changes haven’t introduced any bugs. This leads us to integrate testing into programming, which can act to improve our architecture