I have a stock quote when I'm interviewed about the enterprise service bus (ESB) concept: "The bus does not stop here." My point is that the bus itself is not the be-all and the end all; it's just the enabler. It is the enterprise services that get whatever needs to be done done.

Together bus and services make up an architecture (the acronym SOA is in there somewhere). Dave Rosenberg, CEO of MuleSource, a provider of open source SOA infrastructure and integration software, has a variation on my quote that I like better than my own. Dave takes my architectural analysis and puts it into real-world user terms. Dave says, "Enterprises are deploying multiple ESBs that ultimately develop into a backbone." His point takes my point one step further: it isn't a single bus that is important but multiple buses working with multiple services.

In addition to CEO, Dave Rosenberg is also a founder of MuleSource along with Ross Mason, chief technology officer. Prior to joining with Ross, Dave served as Chief Information Officer for Glass Lewis & Co., an investment research and proxy advisory firm where he used the Mule open source software (OSS) code. Earlier, Dave was Principal Analyst for the Open Source Development Labs (before it merged with the Free Standards Group to become the Linux Foundation); it was there that he received his introduction to OSS. More about Dave's and MuleSource's OSS bonafides below but follow Dave's thinking on infrastructure and integration software functionality vis a vis SOA and the SOA market first.

Rosenberg says, "One person's SOA may be someone else's Web Services, EAI, ESB or other acronym."

The big question enterprises should ask is "What is the right SOA approach for my environment?" At the highest level, the fundamental challenges are the same.

• "It's tough to get data to and from different systems in different formats.
• It's tough to migrate large volumes of data from legacy systems onto new systems without service disruption.
• And it's tough to juggle the ongoing churn of legacy applications and new services that need to be wired together."

That's the functionality part of the story.

As for the market side, based on my work at Research 2.0, I estimate the market for integration and messaging infrastructure software at around $9 billion in 2006. That does not include any associated implementation, hosting, training and other one-time services, which are often two to three times as much based on user spending surveys. Very little of this software would meet the purists definition of SOA today but eventually there will be a total cutover to SOA integration and messaging infrastructure software (just as there has been a movement to client/server software from monolithic software over the last 20 years). Today, only a few suppliers reap the majority of that $9 billion in revenue. They include IBM with approximately $3 billion; BEA and Oracle at about $1.5 billion each; and Microsoft, SAP, Software AG, and TIBCO reaping close to $.5 billion each. As the next cutover takes place, there is an opportunity to replace any or all of the suppliers mentioned above.