I recently wrote an editorial as editor-in-chief of .NET Developer's Journal in which I openly questioned the value of re-architecting existing systems to use the latest and greatest technologies. Specifically, I illustrated my argument with the case of a local ISV (independent software vendor) I know that spent several months re-architecting a successful COM-based application to use .NET while its competitor continued to add features to its existing system. The end result was that after a few months of development by both organizations, the company that had stayed on its legacy platform and invested in features had a much richer product than the folks who had spent their resources porting to .NET.

Unfortunately, this editorial has been misunderstood in some quarters as an attack on the virtues of upfront design versus most organizations' "just start coding" approach. Others among my critics have suggested that I bring the value of the .NET platform into question whenever I suggest that not every legacy system out there is an equally good candidate for migration to my favorite platform. To both of these groups, I have just one thing to say: get real!

As a Microsoft Most Valuable Professional for the .NET Compact Framework and principal consultant for Magenic Technologies, Inc. (http://www.magenic.com), a premier software development and consulting company focused on Microsoft technology, I am well acquainted with the value of up front design and education. Organizations that refuse to do proper planning and design before embarking on the construction of any modern, large-scale system will soon find themselves bleeding money like stuck pigs once they get midway in their development efforts.

This truth will become increasingly evident as the move towards interconnected, service-oriented architectures (SOA) continues to accelerate over the remainder of this decade. SOA has rendered forever obsolete the idea of the lone programmer acting in complete isolation as analyst, architect, and coder for any serious software project.

We are fortunate as software engineers to have a platform as robust and extensible as .NET upon which to build the large-scale SOA systems of the near future. Virtualized execution on the CLR (Common Language Runtime) represents the same quantum leap forward in reliability and security as the move from DOS to Win32 before it. The ability to transparently access, store, and expose object-based functionality via industry-standard XML is - although not strictly required by SOA - so fundamental to 21st-century interoperability requirements as to make any attempt at modern software development in its absence nearly impossible!

The devil in all of these details, however, is the phenomenal difficulty that can arise in implementing any of these technologies without sufficient expertise. .NET reduces the knowledge base a developer needs in order to be successful with SOA to the absolute barest minimum. It is for this single reason that I expect .NET to become an increasing dominant platform as the role of SOA within organizations continues to expand in 2005 and beyond.

In preparing this editorial for publication in WSJ, I consulted a number of software engineering experts here in Illinois, where I live. Arguably the most famous person to respond was Ralph Johnson - one of the coauthors of Addison-Wesley's famous Design Patterns book and currently on the faculty at the University of Illinois at Champaign-Urbana. To read about what he had to say, as well as my response, visit my blog at http://derek.blog-city.com/.