Information Technology Reference
In-Depth Information
time, the promises of MDE have been elucidated for the development of new
applications, providing modeling tools, development tools, domain specific
languages, and the like. A virtual plethora of standards and applications have
been created to support the development of new model-driven applications,
however little has been done to address the cost effectiveness of leveraging
existing systems in a model-driven environment. Douglas Schmidt [4] in his
February 2006 Model-Driven Engineering overview states “When developers
apply MDE tools to model large-scale systems containing thousands of
elements, they must be able to examine various design alternatives quickly
and evaluate the many diverse configuration possibilities available to them.”
He refers to the Integrated Modeling Approach of Lockheed Martin Advanced
Technology Laboratories as an example of legacy integration with less than
ideal results: “Reverse engineering is used to build models… from existing
source code. Many previous attempts to reverse-engineer models from source
code have failed due to a lack in constraining aspects of interest.” A similar
experience described in this paper.
In his article “The Pragmatics of Model-Driven Development”, Bran
Selic[6] discusses legacy integration mostly in terms of development tooling
only tangentially touching upon the issue of leveraging application source
code in the modeled environment. In this case, the recommendation was to
take advantage of legacy code libraries and other legacy software where
domain-specific knowledge often resides. While certainly true, this view
overlooks the problems of impedance mismatch between the two designs and
often-prohibitive implementation cost of custom glue code.
At the UML Conference of 2003, Jean Bezivin [2] presented “MDA: From
Hype to Hope, to Reality” where it was stated: “The extraction of MDE-
models from legacy systems is more difficult but looks as one of the great
challenges of the future. There is much more to legacy extraction than just
conversion from old programming languages”. Indeed, it is the variability of
legacy systems, platforms and ultimately the model impedance mismatch that
is at the heart of these challenges.
Unfortunately, most work in the Model-Driven Engineering area focuses on
the rapidly accelerating model-driven technologies and patently avoids
dealing with the big white elephant in the middle of the room - how to
leverage the existing application features and functionality in an efficient and
cost-effective manner.
Search WWH ::
Custom Search