Long story short, if you spared the time to read my take on the book "Open Source SOA" by Jeff Davis (Manning, May 2009), stop it now, buy the book and have fun. I'm sure you'll love it from the very first page. It contains 5 parts and 12 chapters, each loaded with plethora of an extremely useful, from-the-trenches knowledge accompanied by 1-chapter-long theory. With its pleasant and steady flow it stands out from the other books I've read. The book's title is pretty humble and I wasn't so sure whether or not it'd been worth its time. There has apparently been no rhyme or reason to picking it up from the bookshelf. I wish myself making such choices every time! Perhaps they were the terms "SOA" and "open source" that caught my attention and I don't regret it whatsoever.
Apart from the part "1. History and principles", each and every part is made up of two, three chapters - one with a theory following by practical one(s). Each feature of the final offering called purposely Open SOA Platform has its own part. To me the author is a real subject matter expert (SME) who has achieved something exceptional that I've seen many planned but fell short, most likely for their lack of deep knowledge. After reviewing available open source solutions, the author picks up the right ones to lay down the foundation for successful SOA adoption with no or very low budget without compromising its principles. SOA is a standard-based, service-oriented architecture, but the book made a slight change in my take on SOA - it's no longer an area for commercial products only, but open source products are as feature-rich, viable and can play so well as their commercial counterparts, if not better (price aside). If you've ever thought about SOA as a toy for big players in a industry, I'm quite certain that the book will quickly change your mind and show how easy and beneficial it is, and open source tools make it achievable. Your doubts about pros and cons of SOA, if you've ever had some, should quickly go up in a puff of smoke. The book shows what open source solution to use, when and how. It's sort of a cookbook of SOA for practitioners as well as newcomers. No matter how much expertise you possess, you will certainly find many invaluable chapters.
The author shares his wisdom in a way of many useful comments and notes. Excellent explanations of likely well-known yet often confused terms helped me a lot. I enjoyed the author's writing style and pragmatic approach very much. He is open-minded, touts and criticizes wisely, without showing his appreciation for open source solutions only. I believe he could write as much about open source- as about commercial offerings not only zeroing in on its price, but their features and usability. Since it's a book about open source SOA the author took it very seriously and kept it to the point. The author describes his own experiences with the products (and their source code!) rather than focusing on the features he could read on their websites. The chapters require undivided attention to fully appreciate the book's quality. Luckily, examples are kept to their minimum to keep your attention focused on important details.
Although the first part is meant to introduce the concepts and open source solutions, their quantity and coverage can easily cause a headache. Suffering due to very detailed subject study is not something far away while reading the book. The part 2. "Assembling components and services" is all about SCA and SDO with Apache Tuscany (with some reference to Spring Framework as a IoC/DI framework). It's obvious that the author believes SCA and SDO have their role in enterprise development. The following part 3. "Business process management" touches on, how the author describes, "a BPM solution as rich in functionality and as stable as JBoss jBPM" (page 123). It's all I could've wished for about JBoss jBPM which I had played a bit before. The part 4. "Event stream processing, integration, and mediation" was a completely new part of enterprise architecture with Esper and Apache Synapse. Apache Tuscany, JBoss jBPM, Esper and Apache Synapse begun their play as one. I wish the chapter about Apache Synapse had been written as two parts - one introductory and the other with configuration details. The last part 5. "Enterprise decision management" was exactly on time and with its demonstration of JBoss Drools (aka JBoss Rules) let me compare it with its commercial alternative I just begun my endeavor with - ILOG JRules. I had troubles understanding the vocabulary surrounded BRMS, but the book made the trick and completely removed my confusions.
Despite all the immense SOA learning, I also learnt a lot of new English words that I have never come across before, although now I couldn't believe I did not. They include to premise, an upshot, to preclude, to craft and these are only a few of the plethora of the newly-acquired words which I wrote down from just a single, first page. The book's written in clear, comprehensible English any foreigner will fall madly in love too.
There is one thing I hope will be corrected in the following revisions - it's the use of the words 'I' and 'we'. The author uses 'I' throughout a couple of chapters and out of the sudden does he change his perspective to 'we'. Switching back and forth happens quite a lot and can cause pain for picky readers :)