It is quite common these days to hear someone saying that enterprise architecture has failed to provide expected value. Some people declare that EA is dead. Not that anyone agrees with that, but you can’t deny that certain degree of disillusionment does take place. Practitioners are increasingly disappointed with apparent lack of focus of ‘classic’ all-your-system-are-belong-to-us EA frameworks (read TOGAF) and desperately trying to find a leaner, more value-driven way of doing things.
This book is essentially about post-framework EA. It isn’t new, in fact it was published 3 years ago, but, like a good wine that gets better with age, the book becomes more and more relevant with each passing year. Roger Session courageously redefines what good enterprise architecture is about. In his view, there’s a single fundamental root of every notorious issue in modern enterprise IT such as business misalignment, untimely and unreliable information, soaring costs. The root is complexity that poisons everything if not properly controlled. The goal of this book is to present a proper set of thinking tools that enable an architect to understand, measure, plan and contain the overwhelming complexity of an enterprise.
First few chapters are dedicated a theoretic introduction to the subject of complexity. Author draws from combinatorics and set theory to illustrate several techniques to battle complexity: partitioning, simplification, and iteration. The less math-savvy you are as a reader, the more challenging the reading would be, but at the same time the more you get from the book. Anyway it’s not a rocket science, so everyone could make sense of it.
The rest of the book is mostly dedicated to author’s own lightweight method on controlling the complexity of enterprise architectures which is called Simple Iterative Partitions (SIP). From its point of view an enterprise is seen as a hierarchical composition of autonomous business capabilities (ABCs) that partition both modules of business and software systems that support them. A nice feature of SIP is that it draws a clear line between enterprise and solutions architecture and doesn’t attempt to get into the domain of latter, but rather helps to define boundaries of systems that need to be developed or acquired.
In addition to that there’s an entertaining case study of complexity (mis-)management given on an example of notorious multi-billion atrocity of NPfIT programme that is run by British National Healthcare System since early 2000s. A superficial application of SIP demonstrates that certain programme-level decisions made in the past were fundamentally wrong from the complexity control perspective, and a better way of achieving the goal is elicited. Of course devil is in the detail so I wouldn’t take the conclusions deadly seriously; however, for a sake of method demonstration it is really valuable perspective.
I found the book to be easy to read. In fact it is quite short, less than 200 pages of actual content that is concisely summarized in the last chapter for further reference. It could be even shorter if author took out lengthy introduction on the subject of EA in the first chapter. It is completely useless to most readers because it presents yet another overview of Zachman Framework and TOGAF that is too short to be practically useful on its own and doesn’t add any value as it is not referenced in the book later.
In summary, I would suggest reading this book to anyone practising EA as it provides a completely unorthodox perspective, which is always good since it feeds your critical thinking. On the other hand if your professional interests are closer to tin, e.g. technical architecture, you may find this book of less interest as it lacks any specifics or technologies. It isn’t a book that you’ll keep at your desk, but it is definitely a smart and enjoyable reading that changes the way you look at everyday things.