Convergent Architecture: Building Model Driven J2EE Systems with UML

Author: Richard Hubert List Price: $39.99 Our Price: Click to see the latest and low price ISBN: 0471105600 Publisher: John Wiley & Sons (15 November, 2001) Edition: Paperback Sales Rank: 94,825 Average Customer Rating: 4.33 out of 5

Buy now directly from Amazon.com -

Customer Reviews

My favorite quote from the book;

What large IT organizations need is less philosophical discussion regarding absolute truth and more agreement on Architectural Style.

Another great theme that I'm kind of parapharasing;

If we built airplanes like we build software there would not be much of an airline industry. Most of that is due to style, no one builds an airplane from first principals, why do we so often build software that way.

This book is worth the money. Like one of the other reviewers I was dissappointed in the 'tutorial' nature of everything past chapter 3. However, I thought the first 3 chapters were worth the price and more importantly the time to read it.

For example in Chapter 1 the author goes into the 'higher level of communication' we get out of a style. When someone says 'car' a huge amount of information is transfered in that small 3 letter word. With an architectural style, the author argues, we can begin to have this same level of high bandwidth communication.

Hopefully we have all been on at least one project where there was a practical mind meld in the team. All to often though we are on teams that have to have 2 hour meetings to define the word 'entity'. Until we can perform mind melds for real a 'style' will help us to communicate better.

I give four starts instead of five due to the tutoral nature of the last half of the book.

There is a slowly-growing body of experience in proven approaches to high-efficiency software development. This book shows Richard Hubert to be clearly among the front-runners in the field.

A foundation observation in this field is that, across many projects, there is often much commonality. "Silo" or "stovepipe" developments ignore this commonality. A core goal is to capture commonality and apply it as widely as possible. This is feasible because the technical content of many projects is similar: one web service app is broadly much like another, as are most mainframe batch apps, or stand-alone GUI apps. However, capturing commonality alone is not sufficient. Other key factors are architecture, process, tools, and organization. They must all be brought together into a cohesive whole.

Most industries (other than IT) have known for many years that meeting their "time to market" goals requires this kind of cohesive approach, which Hubert calls "Architectural Style", and fully describes at the start of the book. Thus designing an building a range of products that are similar (such as a range of cars, or a range of houses, or a range of hairpins) can benefit from having the same architectural style, which is understood across the industry, is taught in schools, which evolves over time, and which implies a given set of structural concepts, processes and organization, tools, and technology outlook. Hubert explains how these are the four major elements of an architectural style.

The particular architectural style presented in the body of the book is the "Convergent Architecture" of the title, and addresses distributed applications such as web services. This style has as its metaphor not the mass-production of Chaplin's "Modern Times", but rather a machine shop with highly skilled engineers, each doing what they're best at. Hubert presents the Convergent Architecture in terms of its major elements - base enabling concepts (which he calls the metamodel), component architecture, the IT organization model, development process, and tools. Along the way, project design and management is covered. Also, the process of taking components from the business model straight through to implementation shows how one of the major promises of OO can be delivered. And the way it's done-this is what MDA really means!

It might be thought that Hubert's approach requires a revolution, but this is not so. Like other practitioners in the field, he proposes an evolution, and includes process support for its management. For example, he shows in some detail how RUP is tailored for the Convergent Architecture, so building on currently-available products. He also touches on some of the low-down and difficult challenges, such as how much technical glue to write to make life simpler for application developers - the more you write, the further from standards you get.

Last, but not least, there is a worked example, using Hubert's company's Convergent Architecture IDE, "ArcStyler", from business model right through to EJBs. Now it might be thought that spending over 40 pages on a proprietary product is mere publicity-seeking. But it's not. It's proof that the tool market is now starting to address one of the great challenges facing our industry: how to move away from the cottage industries of stovepipe development into seriously productive development. And if it also advertises the product, well, if it does the job, what's wrong with that?

Finally, this book is well-written, well-structured, and at 250 pages, a comfortable read. If you've read any of the few other books in this field (such as Peter Herzum and my "Business Component Factory", or "Software Product Lines" from the SEI), then this book is an essential companion, bringing valuable new insights. If you haven't, then this book is not only a first-rate introduction to the field, but also a major contributor to it.

Similar Products