It seems integration and SOA are intertwined for most companies, despite what the experts say about the connection being overhyped. You just can't escape the integration value proposition of SOA. Maybe it is EAI Redux -- I don't know. But over and over again, I run across something new about SOA and integration.

The most recent example was this post by Chief Technologist and IT Toolbox blogger Eric Koch. He references a July CIO Insight feature on Valero Energy's SOA deployment. Valero is North America's largest refiner and, according to the article, integration problems drove the company to adopt a service-oriented architecture.

Koch pointed out that Valero focused on integration, which he called "a SOA entry point I have had a great deal of experience and success with as well." And for you SAP fans out there, you'll find that SAP played a starring role in both prompting the energy company to move to a Web services-based SOA and in deploying it.

The big problem Valero faced -- and this may well be familiar to many of you -- was with proprietary interfaces. With each acquisition or system change, Valero had to rebuild its interfaces -- and each application included about 10, according to the company's CIO. In case you've never run the numbers, here's what the article said about the costs:

"It costs about $50,000 to $100,000 per interface for every five to 10 years of use, according to industry estimates, in addition to the people, time and money to support each interface, update it when there are system changes, and store duplicate data moving between systems."

IT spent all its time fixing broken interfaces, which just kept breaking. Valero decided it was time for a change. But it also looked farther ahead while deploying its SOA, to a time when business users would be able to build their own software.

What's really impressive about Valero's SOA deployment is it managed to demonstrate that service reusability can happen with SOA, even though reuse as a SOA value proposition has fallen out of vogue. Valero managed to hit 50 percent reusability of services and believes it will hit 70-80 percent reuse in the future.

Valero's SOA implementation demonstrates that SOA can help resolve some recurring integration problems in large companies with legacy systems, which has long been seen as a value proposition.

But in a recent CIO.com column, IT veteran and chief architect Mike Kavis suggests SOA could be useful to young, start-up companies as well.

Obviously, start-ups don't have legacy systems and the related integration problems, a fact Kavis' title acknowledges: "Startup SOA Because It's More Than Integrating Legacy Systems."

On the other hand, start-ups don't have the cultural and technology barriers to SOA that established companies face. By opting to use a service-oriented architecture from the get-go, Kavis argues, start-ups can avoid future integration problems, make it easier to hook up with SaaS offerings, and plug into the systems of future customers and partners.

Plus, he points out, start-ups probably won't need the expensive middleware and governance systems you typically need for large company SOAs.

In short, Kavis said it's easier to do things "right" when you're starting out than to overhaul an existing infrastructure:

"As the larger corporations struggle with large scale SOA initiatives, startups can make SOA part of their IT strategy from the inception and take a 'built to change' approach to software development.... when it comes time to deliver reliable production versions, startups should consider implementing a light-weight SOA to achieve the long-term benefits of reuse, agility and flexibility."

SOA is definitely in the trough of disillusionment, but these pieces make a strong case for SOA's long-term integration value.