Skip Links

Network World

  • Social Web 
  • Email 
  • Close

(Comma separation for multiple addresses)
Your Message:

Making sense of service-oriented architectures

Eye on the Carriers By Johna Till Johnson , Network World , 10/25/2004
Johnson
  • Share/Email
  • Tweet This
  • Comment
  • Print

Afew weeks back I attended the Vortex conference, which features industry luminaries discussing tech trends for the enlightenment of those of us who need to plan strategies and budgets. The focus this year was on service-oriented architectures, which, if you believe the pundits, will revolutionize software and networks over the next three to five years.

IT executives might be forgiven for having a smidgen of skepticism: How many times have we been told that X technology is "revolutionary"? How often has that been the case? This time, though, I think the talking heads have it right.

A bit of background: SOA is the latest answer to the perennial problem of reducing the complexity and interdependencies of componentized software systems. SOAs today are built on Web services, delivered primarily as Simple Object Access Protocol/XML interfaces and Web Services Description Language definitions. Like objects and software components, "services" are fundamental building blocks that can be used to construct complex systems across a distributed network.

A simple way to look at SOA is that it represents the true commercialization of peer-to-peer computing. Ultimately, SOA provides a framework for software on one system to securely and reliably request and receive computational resources on other systems - without requiring a centralized server to manage and administer the entire network end-to-end.

An analogy might make it clearer: Think of the network evolution from SNA to IP. In an SNA world, the architecture is basically "master-slave": A gateway controls connectivity between remote terminals and the host, with intelligence residing primarily at one end of the connection (the host). With IP, a distributed network of routers provides the connectivity. Control is no longer centralized; it's shared.

However, software architecture hasn't entirely kept up with the IP revolution. In so-called client-server computing (the dominant paradigm today), the fundamental structure remains master-slave: Application servers control communications with remote clients.

Over the past few years, we've tweaked the client-server model slightly around the edges. Communication protocols between clients and servers are increasingly Web-based rather than proprietary, and applications have become multi-tiered rather than single-tiered.

  • Share/Email
  • Tweet This
  • Comment
  • Print
Comment
Login
Forgot your account info?
Add comment
Anonymous comments subject to approval. Register here for member benefits.
Have a NetworkWorld account? Log in here. Register now for a free account.

Videos

rssRss Feed