Date of Award
Master of Science in Computer Science - (M.S.)
Computer and Information Science
Vassilka D. Kirova
Fadi P. Deek
Phillip A. Laplante
The movement of distributed applications from 2-tier to n-tier architectures have enabled systems to be scaled to meet the demands of an ever increasing population of users. Two middleware architectures have come to the forefront: Microsoft's DCOM and the OMG's CORBA. These are not the only possible architectures for n-tier distributed applications, but they are currently the only two which offer a degree of platform independence and the flexibility of using different programming languages for development.
CORBA provides platform independence because it provides a middle layer between the client and the server and services client requests using its internal naming service to identify server objects and then expose methods to the client through it's object adapter (POA). CORBA is a self contained middleware that operates independent of the underlying operating system. CORBA offers the potential of ease of maintainability since server objects can be changed and the new methods can be discovered at runtime by the client using CORBAs Dynamic Invocation Interface. Client code would therefore not have be recompiled as it would using static IDL mappings and client and server stubs.
DCOM, in contrast is a platform dependent solution that can only be used on Windows machines, although ports for other platforms are in the works. It relies on the Windows registry to identify objects and the operating system to assist in runtime control of objects. Because DCOM is nothing more of a remote extension to the already established Common Object Model which all contemporary Windows operating systems and applications are built upon, it may provide the easiest path to distributed applications for Windows developers that are already familiar with the Common Object Model.
Reynolds, Reginald J., "Comparison of DCOM and CORBA distributed computing" (1999). Theses. 874.