First Name : Edward
Last Name : Cobb
Title : Technical Director
Company : BEA Systems, Inc.
Address : 2315 N. 1st St.
City : San Jose
State : CA
Mail Code : Telephone : 1-408-570-8264
FAX : 1-408-570-8942
E-mail : firstname.lastname@example.org
Cyberpaper : Server Side Component Models, the future of business computing or just another technology fad. It's hard to pick up a trade magazine these days without coming across an article on one or more of the leading server-side component models. Beginning with Microsoft's MTS through Sun's announcement of Enterprise Java Beans at Java One in 1998, everyone is talking about the promise of component models for server-side development.
The most recent addition to the list is the OMG effort to produce a component model for CORBA which will extend the EJB model to multiple languages and define a container based on CORBA infrastructure. The CORBA model not only has features from EJB and features which exploit the CORBA infrastructure, but also has features which can be traced to Microsoft's COM. Multiple interfaces with navigation between them, an event model that relies on a reliable queuing engine for distribution, and assembly from parts give the CORBA model the potential to support all the other models and provide an integration of both EJB and MTS programming styles in a single server environment.
Even if these claims were true from a technical viewpoint, the reality of achieving such goals in the commercial marketplace is something quite different. In fact, the growth in the EAI market based on simple messaging technology calls into question the very premise of the component model advocates, viz. that a server-side component model is the critical element of an application server platform for the next decade.
In the end, the things that have made a difference in TP systems from the beginning (reliability, performance, and availability) may have as much to say about the outcome of this debate as the technologies themself. I believe that the component model has the best potential, in large part because much of the mundane aspects of application development can be handled by tools and the ability to customize reusable components rather than develop them can significantly reduce the development cycle. I would further argue that the EJB/CORBA alternative is better suited to the diversity of today's IT environments. But that apparent strength is also its greatest weakness. Our industry's track record on cooperation is not good. Individual business interests always seem to get in the way.