Dismantle the overwhelming complexity in your IT projects with strategies and real-world examples from a leading expert on enterprise architecture. This guide describes best practices for creating an efficient IT organization that consistently delivers on time, on budget, and in line with business needs.
IT systems have become too complex—and too expensive. Complexity can create delays, cost overruns, and outcomes that do not meet business requirements. The resulting losses can impact your entire company. This guide demonstrates that, contrary to popular belief, complex problems demand simple solutions. The author believes that 50 percent of the complexity of a typical IT project can and should be eliminated—and he shows you how to do it.
You’ll learn a model for understanding complexity, the three tenets of complexity control, and how to apply specific techniques such as checking architectures for validity. Find out how the author’s methodology could have saved a real-world IT project that went off track, and ways to implement his solutions in a variety of situations.
Sessions discusses some of the key problems in enterprise architecture and identifies complexity as the primary obstruction to resolving these. He uses some fairly simple maths from probability & set theory to both characterise the problems and as basis for his proposed method of tackling the issue - iterative partitioning of the domain into "autonomous business constructs" and strictly defining the boundaries and interfaces between them.
As one of his prime examples, he dissects the NHS' National Programme for IT, identifies the inherent complexity at its heart, and predicts its total failure. Which admittedly, could be said of pretty much every IT professional in Britain who was not working on the project, but his explanations of exactly why the project's attempted risk mitigations actually made matters worse, while increasing complexity at the same time, are well worth reading (and should be mandatory for anyone commissioning large government IT projects)
Throughout, he make his case well, and his methodology is likely to lead to less complexity in the enterprise architecture, but reading the latter half of the book, I couldn't help but feel that there's not actually a lot new here - for "simple iterative partitioning" and ABCs, read domain-driven design and a refined class model, albeit taken to a somewhat higher level of abstraction.
A very clear book, focusing on a method to create simple architectures.
The author notes that Zachman provides a set of artefacts, but doesn't say how to create them. TOGAF provides a process to follow when creating an architecture, but doesn't describe how to create the architecture itself. And FEA provides ways to judge organizational adherence to an architecture.
The book describes a method that complements Zachman, TOGAF, and FEA, or that can be used stand-alone. Simplicity is valued above all else, since complexity can drive up implementation costs exponentially.
It is a solid book. I enjoy the chapters on complexity calculations and the definition on complexity. It isn't junior friendly (this isn't a bad thing, just saying) but it carries a lot of value and gives a good introduction to the world of enterprise architectures.
My main issue with this book is that terminology in this book isn't google friendly. When I googled some of the definitions/names the only relevant results I got were from links pointing back to the book. This does raise some questions and also makes the book relatively hard to compare to other books/materials on the matter.
I felt this was a great book by Mr. Sessions. It is important to remember that simple architectures are easier to use and maintain long term. This usually also leads to best performance and best overall satisfaction.