It could be called the yin and yang of system design or maybe the challenge of developing a system containing two opposing forces. While application programs are becoming dispersed chunks of content that are assembled on the fly for specific business purposes, the wide-open Internet on which those applications reside is being partitioned, segmented and gated. Programmers like big, difficult problems, and the current state of information system design is about as complex as it gets.

At one point it seemed as if information system design would resolve itself into Java programmers, .Net programmers and folks building bridges between the systems. But along the way to programming simplification, systems built on those widely accepted languages are running up against the real world.

In that real world, the Internet is quickly being transformed from a wide-open (the current popular term is "flat") system to a walled, segmented system. The most visible example of this segmentation on a big level is in China. In a move ostensibly designed to defeat online pornography, Chinese officials are building a system that can filter content and deny access based on whatever parameters they set. The system uses technology products from well-known vendors such as Microsoft and Cisco and has touched off cries of injustice from critics contending vendors are putting profits ahead of democratic ideals. While examples such as that in China are getting widespread—and deserved—attention, many technology executives are deep into deciding how to maintain open access to customers and suppliers but restrict access in accord with government regulations and good security practices.

"We're open now, but I see a lot of companies really controlling how information comes in and goes out. If it gets too chaotic out there, we may lock ourselves back more," said Kevin Wilson, an eWEEK Corporate Partner and Duke Energy's product line manager of desktop hardware, in a recent eWEEK article. "We're also starting to pull systems off the 'big I' Internet wherever possible ... especially if systems have student or employee information on them," agreed Kevin Baradet, at the S.C. Johnson Graduate School of Management at Cornell University, in the same article.

In the meantime, the move to SOAs (service-oriented architectures) continues forward. Sun Microsystems, Novell, IBM and Hewlett-Packard are only a few of the vendors that have recently introduced new or expanded SOA products. In the SOA model, applications are morphed into business functions (human resources, supply chain management, customer management) and assembled as the need requires. This makes sense when you develop SOA networks to operate on internal networks; it becomes more difficult when you start to tie in supplier systems, and it becomes a real sticky information architecture problem when your SOA system for, say, supply chain management has to access your suppliers in China, where the content will be filtered. Better hope those power supplies you use aren't sold by a company named Democracy.

The advocates of SOA (include me here) view these new applications as a way to continue the productivity push that is the basic engine of economic development. However, as with many new technologies, the problems are often understated. What has been missing from the discussion is how issues such as security, regulatory compliance and now government censorship will be accommodated at the same time broad, globe-spanning applications are developed. Those business functions that are the heart of SOA applications are created on a real-time basis drawn from databases and application servers from many countries. The failure of any one of those systems to produce the data required or the sudden change in a government policy to restrict data that was once available calls into question the integrity of the entire system. Vast interconnectedness is great when it works.

Here in the United States, the loss of technologists who can figure out this intricate information architecture puzzle to downsizing, offshoring or disillusionment may indeed hinder the future of global enterprise computing.

Editor in Chief Eric Lundquist can be reached at eric_lundquist@ziffdavis.com.

This article was originally published on eWEEK.com.