By Roger Sessions
208 pages
http://www.amazon.com/Architectures-Enterprises-PRO-best-Practices-Microsoft/dp/0735625786

I saw this book on the shelf in the Microsoft store at TechEd 2008 a couple weeks ago, and when I read the title I seriously laughed out loud.  I thought, "Finally ... the answer to all my problems, and they were able to fit it in just over 200 pages.  Amazing!"  I didn't even pick it up.  But, during the 2nd week of TechEd I was hanging around wasting time between sessions and curiosity got the best of me.  I wanted to know just what kind of b.s. could possibly be between those covers.  Within 5 minutes I bought the book, and after I started reading it I actually ordered another copy off Amazon and had it overnighted to a developer I work with so he could read it too.  Every now and then I read a book, and go "holy crap, this is really useful stuff" ... and this is one of those books (Code Complete, and Don't Make Me Think are others).

Roger Sessions outlines why the top issues CIOs claimed they faced in a December 2007 article in CIO Magazine (e.g. improving IT planning processes, improving project management capabilities, reducing IT costs) weren't actually problems ... they were side-effects of a fundamental problem of the high complexity inside most organizations.  Most enterprise architectures are extremely time consuming and expensive (not to mention critical because are the foundation of an organization's entire technology infrastructure).  But Sessions exposes the truth that most enterprise architecture decisions today are based on instinct, gut feeling, politics, vendor loyalty, and so on, and because of this the only way to know if an particular architecture is good or bad ... is implementing it (a very big gamble).

Sessions looks also looks at many of the popular enterprise architectures used today, and doesn't seek to replace them with some revolutionary idea, but instead he tries to compliment them with some additional methodologies he called SIP (Simple Iterative Partitions) which gives a special focus to the root cause of so many IT project failures ... complexity.  He doesn't just present his opinion on how enterprise architectures should be done, but instead performs thorough, scientific, mathematics-based investigation on how to make an architecture as simple and modular as possible.  In the end he provides some practical steps on how to integrate this thinking into your current infrastructure.  Like so many new trends, he doesn't expect you to (as Scott Hanselman jokes) open a new window, and go to "File > New Company" ... but suggests a realistic, iterative approach for how to work this in as a fundamental philosophy of your systems without shutting down the IT team for a couple months to work this stuff in.

Seriously, this is probably the most useful book I have read so far this year.  I am going to take its message to heart, put it to work immediately, and I know my business will experience more value, flexibility, and stability because of it.