Many in the media are already calling on the demise of SOA and saying that it's just a passing phase, or it's really just a rebrand of the EAI space, or that it will be segmented only to certain integration-type challenges.

We have a bit of a different take. The term SOA will go away over the next several years but it will go the same route that e-commerce applications went.

If you think back to the mid-'90s, we started talking about e-commerce as an architecture that was distinct from the then traditional enterprise architecture. However, over the past few years, e-commerce as a term has disappeared because it has become the ubiquitous expectation for how we build applications, or what those applications do (some kind of commerce). It consumed the entire enterprise application space.

To pick one SOA platform - we used to buy BEA for the Website part of our applications, and then kept building in a completely disjointed fashion inside the firewall. Today, however, that same BEA server is used both inside and outside the firewall. This is probably an e-commerce implementation as well; we just don't call it that anymore. SOA will become the same kind of default for enterprise applications.

SOA is a new way of looking at a very traditional, classic problem, so I don't think of it as brand new. At the same time, I don't think of it as just a re-named EAI or some passing fancy. In fact, SOA is the way to look at application design so that the business can understand it and it can be better decoupled from all of its interdependencies; as such, the practice of SOA will live on.

Feel free to stop calling it SOA in a few years, when that's just the expected way we build applications. Feel free to not call it SOA at that point, because you will be doing it de facto. Is SOA dead, doomed, or misnamed? It's none of those, but we have to put a name on a new idea until it becomes old hat - and SOA is that name.