 |
|
Enterprise CORBA, by Dirk Slama, Jason Garbis, and Perry Russell, is part of a new generation of books about CORBA -- books that are no longer about learning CORBA, but doing it. The book provides the experienced engineer who already possesses a basic grasp of CORBA with an overview of issues surrounding the development, deployment, and maintenance of enterprise systems based on it.
Enterprise CORBA is not so much about doing things in CORBA as it is about doing things in CORBA that are not easy to do, but that have to be done to make a project deployable. This is refreshing. CORBA technology is deep enough that most books concentrate on what CORBA does and does well. The authors of Enterprise CORBA clearly have passed numerous evenings sweating over that which CORBA does not do well or easily.
The table of contents (see below) is enough to reassure the experienced distributed-systems designer and implementer that the authors genuinely know what the issues are. What then remains to be seen is whether the authors have anything to say about these issues worth reading.
Well, last time you planned the naming layout of your distributed objects, did you fully consider the implications of the naming hierarchy you chose? Did you consider flat name hierarchy, deep name hierarchy, compact naming, descriptive naming, and federated naming? This is a local manifestation of one of those recurrent themes of computer science, and it's a pleasure to see the authors have latched on to its significance and don't just brush past it with generalities.
On the other hand, I wonder what they meant by "The secret of succeeding with a billion-object CORBA system often lies in the right IDL design." "often"? Given the breathtaking scope of this assertion, more is required of the authors than the two paragraphs that comprise their "Billion Object CORBA System" exposition.
Overall, Enterprise CORBA seems to say as much or more just in parts 3, 4, and 5 as the entirety of 3-Tier Client Server at Work (previously reviewed on ERCB) and in a great deal more detail. Of course, the latter work was really about case histories and top-level solutioning, while Enterprise CORBA is a technical reader's overview of a set of software engineering problems.
Again and again, we find that attention to relevant practical detail, as in the the discussion of Concentrator patterns in Chapter 15 on load balancing. I wish I'd had their exposition of this critical concept, as applicable to complex GUI's operating in a single process space as it is to large networked entities, in print when I needed it some years back to whack a certain design team over the head with...
Slama, Garbis, and Russell are IONA consultants. I'm gaining a great deal of respect for IONA while reading the books their engineers produce (see my ERCB review of COM-CORBA Interoperability). Their insights are penetrating and their familiarity with the technology bolsters the reader's own confidence. And there is little or no overt hucksterism or boosterism for the corporate flag.
The authors have also established an Enterprise CORBA web site at http://www.middlewarehouse.com/, which features, among other tools and topics, an open-discussion BBS based on the layout of the book.
-- Jack Woehr
Part 1: Foundations
1. Introduction
2. CORBA Revisited
3. CORBA Services Revisited
4. Overview of a Simple Example
5. Performance Considerations
Part 2: Core Services
6. Object Location
7. Messaging
8. Security
Part 3: Database Integration and Transaction Processing
9. Object Persistence
10. Database Integration
11. Transactions in a CORBA Environment
12. Distributed Transaction Processing
13. User Sessions
Part 4: Scalability Issues
14. Managing Server Resources
15. Load Balancing
16. Fault Tolerance
17. System Management and Maintenance
Part 5: Engineering CORBA Systems
18. Consequences of the Engineering Process
19. Automating the Engineering Process
20. Conclusion
| Readability |
|
| Originality |
|
| Organization |
|
| Accuracy |
|
| Consistency |
|
| Depth |
|
| Timeliness |
|
| Editing |
|
| Design |
|
| Overall Value |
|
Explanation of ERCB rating scale: No stars = unacceptable, 1 Star = marginal, 2 Stars = average, 3 Stars = above average, 4 Stars = exceptional.