Business Component Factory : A Comprehensive Overview of Component-Based Development for the Enterprise

Much of the material in this book is stuff you've seen before -- the words are different or they're used with different meanings or in new contexts -- but a lot of the concepts are familiar. The book does expand its scope somewhat to cover much more of the "development process", resulting in more of a mix of technical and process information than is typical. This is good, as all too often we tend to separate the technology from the process.

Bottom line: Useful information. Nothing particularly new or revolutionary. Could have been a couple of hundred pages shorter. I frequently found myself needing to re-read something or refer to the glossary to re-discover a definition.


Full lifecycle view, business advatage to large grain CBD   May 15, 2001
Ryan Darby (Pretoria, South Africa)
7 out of 7 found this review helpful

Herzum and Sims have taken a brave approach to components, emphasizing the view of components through the development lifecycle, where the same component exists from requirements through design to implementation, with only a different view on it. They do a good job of completing the lifecycle with project management aspects of component development as well.

The book seems high level, but they enter into a lot of technical detail as well, while not getting involved in a specific technology. The book is used by me as a textbook for graduate students, as it covers all aspects in detail but generically.

The advantage of the book is the way in which components are defined. Business Components are large grained, made up of many parts which they define in layers. This leads to a wider view of the concept, and leads to a re-organization of the development process.

The book is structured around an architecture for development, which establishes a production-line approach. This ensures the component concept is bought into throughout the organization.

This is the only book to focus on large grained components, with a pure business advantage, but explained technically. This is and is not a how to book. It is a roadmap for what to do and how to arrange it, but not the specific technology to use.

There is a lot of detail in this thick book, but it is easy to read. Very unique approach, and the only book describing aspects you will not learn elsewhere. Other books only describe the overall concept. This one tells you exactly how to fit it into your organization, down to how to structure teams! The book is very comprehansive, and really does follow the development lifecycle. You will gain knowledge of : components on a business level, a new lifecycle for development that is very tailored to components in business, techniques for developing systems, from individual components to integrating federations of components form third parties, all the other aspects thinner books leave out.


Dispelling Myths, Doing it Right   August 1, 2000
James R. McElroy (Magalia, CA USA)
20 out of 21 found this review helpful

As an OO practitioner and methodologist for the last 10 years, I found the Herzum / Sims book to be right on the money in several regards.

OO has a lot of theoretical ideas which just don't seem to pan out in practice. The Business Component Factory cleary explains why, and shows what really works in the true industrial setting. It is rich in practical advise, and low in BS. Very refreshing for the software practitioner who is frustrated by the OO theoreticians who spout their wisdom from the ivory towers, but have rarely, if ever, had to work on real projects.

Along these lines, the BCF book dispels the OO myth that all classes / objects must be as intelligent as possible, and admits that, in reality, it is often best to have "focus" classes. These classes contain the intelligence of a group of related classes (grouped in a component) and give the advantage of lower coupling for the other classes, and of providing a focus target for process and use case modeling. Hence, Herzum / Sims tie the use case models effectively to classes, then to components.

The BCF book also points out that components need to be "first class citizens" in the UML metamodel, which map from analysis through design into code. As the UML currently stands, packages and (UML-style) components fail miserably in this area. Herzum / Sims show how to get around this deficiency and model and produce large-scale software units (components) effectively.

There is much more to the book than described above, but the above two points emphasize that the BCF book is not afraid to take on conventional wisdom (even the sacred UML), to point out flaws in this "wisdom", and to discuss what really works. Highly recommended, especially for anyone working on large-scale system development.


An excellent guide to successful adoption of Enterprise CBD   May 29, 2000
Mark S Potts (Perth, Australia)
19 out of 19 found this review helpful

This book is the first I have read that really tackles all aspects of what is required for Enterprise Application Development through a CBD approach.

By defining the levels of component granularity and a recursively discrete approach to breaking a business problem down into components and their constituents as finer grained components, the true requirements for CBD are evident and determined. Many books I have read make the same mistake of only discussing development of components at one level (usually what Herzum defines as the distributed component level) and fail to address the many of the aspects of CBD that are not covered by development alone (deployment, testing, management, integration, and a roadmap for the development process and managment of that process through to delivery of a component based system). The book also talks and applies the component levels to the commonly depicted 4 tier architecture and importantly introduces the concept of components needing to be not only strongly typed for internal systems but also strongly tagged (supporting XML based component messaging/invocation) for virtual and extended systems. The coverage of what is required from a Component Execution Environment (CEE) when components are more course grained than simple distributed components is well covered and continues to define the true requirements for a Business Component Execution Environment (BCVM).

The book is a must read for anyone serious about adopting CBD on and enterprise scale. The book goes well beyond the common text available for CBD (that all concentrate on the short sighted development requirements for distributed components in a fine grained component containment model). I agree with another reviewer that for those of us that have been developing systems in EJB, COM+/DCOM and CORBA much of the book covers lessons we have painfully had to learn in developing multiple component based systems that have to inter-operate, but it goes beyond that in looking at what is necessary for component based systems at the next architectural level (one that may well incorporate disparate distributed component models).


Overall Disappointing   March 29, 2000
9 out of 11 found this review helpful

Although I found some useful bits in here, it did not come across as good to me as some earlier reviewers found it. Maybe I missed something.

I did find some useful coverage of some good development practice e.g. the use of layered architectures, components as large grained deployment and distribution units, explicit component artifacts including external description through code, build and deployment, etc. But I did not get much more from these discussions than I already knew.

I did not find the "interoperability protocol" explanation helpful: e.g. "design-time interface" was a very fancy word for a simple list of the artifacts required at design time. The authors do not seem to recognize that these include both things needed for tools to manipulate a component at design or build time, as well as those needed by developers to understand, use, or test a component.

The book sorely needs a clearer high-level view of how components are configured together, rather than the nitty-gritty of proxies, separating functional vs. platform-specific code, etc. I see some links to the never-ending saga of the elusive "business object" track at the OMG.

Overall I was disappointed. I'd say between 1 and 2 "stars". Your milage may vary.