 |
| Strategic Management Series |
Simplicity of design has been for centuries the wholly grail of architects, while software designers seem to situate themselves in opposition with the trend, as they aim using a mix of technologies that usually increase architecture’s complexity (sometimes the many, the newer and fancier, the better). Unfortunately, despite the implied but not necessarily reachable potential, each component added to an information system or infrastructure has the potential of increasing the overall complexity by a factor proportional to the degree of interactions it creates, respectively by the number of issues it creates or allows to propagate through these interactions.
Conversely, one talks about simplicity in IT without stating what is intended by it, and it can mean many things. Quite often the aim is packed within the ‘keep it simple stupid’ (aka KISS) mantra, a modern and pejorative alternative of Occam’s razor. KISS became a principle in software architecture design, and it can mean that a simple solution works better than a complex one, or that pursuing something in the simplest manner possible is usually better. The nuances are wide enough to cover a wide spectrum of solutions, arriving at statements that the simplest choice to make is the most appropriate one to make, thing that’s not necessarily true in IT, where complexity finds itself home.
Starting with the important number of technologies coexisting in integrations and ending with the exceptions existing in processes or the quality of data, things are almost never as simple as one may wish. An IT infrastructure’s complexity is dependent on the number of existing components, on whether they come from different generations or come from different vendor, on whether are deployed on different operating systems or are supported by different service providers, on the number of customizations made, on the degree of overlapping of the data and integrations needed to keep the data in synch, respectively of the differences existing in data models, quality and use. In general, the more variance, randomness, and challenges one has, the higher the overall complexity.
Paraphrasing Saint Exupéry, in IT simplicity is reached when there is no longer anything to add or anything to take away, or in Hans Hofmann’s words, simplicity is reflected in ‘the ability to simplify means to eliminate the unnecessary so that the necessary may speak’. This refers to the features, what a piece of software can do, respectively the functionality, how a certain outcome is reached, which arrive to be packed in various logical aggregations (function point, functional requirement, story, epic, model, product, etc.) or physical aggregations (classes, components, packages, services, models, etc.). These are the levels at which one needs to address simplicity adequately.
To make something simple one must be able either to design a solution up to the detail that there’s nothing to add or remove, or to start with something and remove or things to reach simplicity. Both approaches involve a considerable effort, time, and multiple iterations, however the first approach can easily become utopian as some architectures are so complex that sooner or later the second approach comes into play. Therefore, one needs in general to focus on what seems an optimal solution and optimize it continuously in further iterations. Aiming for perfection from the beginning or also later in the improvement process is a foolhardy wish.
Even if simplicity is hard to achieve, one can still talk about the elegance of a solution, scenarios in which the various components fit together like the pieces of a puzzle, or about robustness, reliability, correctness, maintainability, (re)usability, or learnability. These latter characteristics are known in Software Engineering as (software) quality attributes.
