San Diego – After the opening morning at the Gartner Application Integration & Web Services Summit featuring keynotes with predictions for Web 2.0, Enterprise 2.0 and SOA 2.0, it was good to find a track in the afternoon titled "SOA in the Real World."
Eschewing the Gartner tendency for crystal ball gazing, Massimo Pezzini, vice president and distinguished analyst Gartner Inc., began that track session with a brief history of SOA. He noted that when he started working for Gartner in Europe in 1996, he was already going to IT departments with a paper outlining the principle of SOA. In many cases, he found he was evangelizing the converted.
"Customers were doing SOA then although they weren't calling it that," he told his audience. They tended to use the terms of the 1990s for their projects, calling them client/server. Pezzini said that is the secret few SOA gurus want to let out of the bag: SOA is an update of classic client/server.
If developers were doing SOA 10 years ago, why is it considered the greatest thing since that anonymous baker began slicing bread before selling it?
Pezzini said two things have made SOA the hot topic at analyst conferences. First, the development of common standards based on XML to help facilitate linking services in an application. Second, he said, "Now we know how to do it because of the pioneers from 10 years ago."
He offered some wisdom gained from a decade of consulting with Gartner clients doing SOA.
One of his first points was that for all the talk of the cost benefits of SOA and reuse, it is a hard sell at the executive level. Injecting a little humor, he did an imitation of a boss having listened to an IT manager explaining cost justification: "You're saying that if I give you $5 million today, it will save $10 million in three years. But if I don't give you $5 million, I can save $5 million now?"
While most analysts, including Pezzini, recommend starting with small SOA projects and building incrementally, he said in reality "SOA is only cost justified in major applications" where there is the potential to save large amounts of money.
However, since reuse is one of the ways to save money with SOA, he suggested following the example of Verizon Communications Inc. and form a "Service Chasing Team." He said Verizon has saved money by having a team of IT professionals dedicated to searching through the telco's large SOA infrastructure and identifying Web services that are ideal for reuse and then passing the information on to development teams.
While the Service Chasing Team provides a carrot for reuse, he also recommended that organizations establish discipline and governance processes focused on avoiding the "wild" proliferation of services.
"You will have to have a formal process for building Web services," he said. "You can't just have developers building services when they like."
Pezzini was not the only Gartner analyst offering advice on reuse. In the opening keynote and in a later track session Roy Schulte, vice president and distinguished analyst, warned IT managers not to over promise on reuse. Because services remain domain specific reuse will only work within "limited domains," he said.
Noting that while there are XML-based standards for Web services, different vendors execute them in "different dialects."
"There is really not a standard Web service," Schulte said. He noted that the old definition of "orthodox" Web services based on SOAP, WSDL and UDDI is being challenged by other "Web-based services" using technologies such as Ajax and Plain Old XML (POX).
As with cost justification, Jeff Schulman, group vice president for Gartner, warned IT managers in the second keynote of the morning not to hang their hat on reuse when trying to sell SOA projects to executives.
"Reuse is an IT issue," Schulman said. He argued that SOA only has value to executives as a way of transforming business operations to meet the demands of the rapidly evolving Web 2.0 business model pioneered by Amazon, Ebay, Google and Yahoo.
While talking about the transformative nature of Web 2.0, Schulman offered an alternative to the maligned SOA 2.0 terminology. He suggested "Advanced SOA" might be the term for future SOA implementations moving beyond request and reply.
For developers who fear that they will not get to do SOA because they cannot justify the costs even by promoting reuse, Pezzini was reassuring.
"In five years everybody will be doing SOA," he said. Moving from history to prognostication, the analyst said that by 2010 SOA will be built into packaged applications from major vendors including IBM, Oracle and SAP.
"Whether you want it or not," Pezzini said, "everybody will be doing SOA because of the packaged SOA."