In the beginning of the 2009, SOA was "dead," the topic of obituaries - by the end of the year, SOA got its very own manifesto. In between, software architects continued to implement SOA, learning all the time. As 2010 begins to dawn - driven in many ways by highly touted cloud computing - the SOA conversation continues and experts are taking note.
Early in January, Burton Group and SOA Practice VP Anne Thomas Manes shook things up when she declared "SOA is dead." She argued that vendors promised SOA would be a way for enterprises to save time and money while modernizing their infrastructures. These benefits, she said, were not realized in real-world implementations.
The truth is SOA can be quite time-consuming and costly to implement from scratch. It is a methodology, not a product, and successful implementation requires a long-term approach.
The US Coast Guard reached a major SOA timeline checkpoint in 2009 when it finished a SOA implementation for its Long Range Information and Tracking (LRIT) system. With an enterprise service bus (ESB) crunching 50 million messages a day, LRIT keeps detailed analytic data on the 6,000 or so ships along the U.S. coast at any given time.
But deploying a SOA was a time consuming process, taking about 2.5 years to get to that point from a legacy system. The implementers took it slow and steady. "We ended up with a completely different technology infrastructure than conventional wisdom dictates," said Steve Munson, chief of enterprise engineering at the USCG operations systems center (OSC). "It's a large shift for your mental approach to things and it can be a big investment."
Shift from throwing code at problems
The shift Munson mentioned is one many SOA implementers face. Instead of data sitting in silos waiting to be referenced by stand-alone applications, the SOA approach requires a set of Web services be built and arranged to share data and be re-used across the system. Throwing new code at every problem can be counterproductive; though this is what many traditional developers have been trained to do.
But despite the hurdles involved, 2009 was a strong year for SOA. In May, Forrester Research released the results of a survey that showed the IT industry is taking SOA very seriously. Out of 2,227 IT executives only 1% of current adopters said they have seen little to no benefit from implementing a service-oriented architecture. Moreover, it showed that 75 % of IT executives and decision makers at Global 2000 organizations planned to be using SOA by the end of 2009.
Some say SOA's pulse quickened this year as a result of all the reflected interest in cloud computing.
"Cloud computing came in at just the right time when SOA was losing its panache," said Dana Gardner, principal analyst, Interarbor Solutions. "The good news for SOA is that SOA is in many ways a requirement for moving to the cloud in any meaningful way."
If an organization wants to utilize cloud computing beyond simple pilots and get enterprise-level control and management, SOA is the top methodology for doing so, Gardner said.
Then in October, a group of industry experts unveiled the SOA Manifesto at the 2009 SOA Symposium in Rotterdam. Based on the previous Agile Manifesto, it preached business value over technical strategy, shared services over specific-purpose implementations and evolutionary refinement over the pursuit of initial perfection.
SOA Manifesto: More blog fodder?
The manifesto plainly codified what many had been saying about SOA all along. But it was received less than enthusiastically.
"Typically manifestos like that are good for press releases and they're good for blog fodder," said David Linthicum, expert and author on SOA and Cloud Computing. "But they typically don't do much good. In looking at the manifesto, a lot of that stuff is common sense in how people want to leverage service oriented architecture."
For a long time the SOA discussion was something driven more by punditry than any critical thought leader, Linthicum said. There was a lot of hype and slinging of buzz words. SOA is not an easy undertaking and can require a lot of training and even the replacement of IT employees. But, he said, at the end of the day it is a valuable undertaking and a great way to get an enterprise "cloud-ready."
Author and enterprise architecture expert Leon Kappelman said the major take-away from the SOA Manifesto is the common necessity to remind IT that its applications and services all need to provide the business with clear value. And that requires good communication.
"I think as long as we use the word SOA the business is going to think we're talking geek and toss us out," Kappelman said. "Whether we call it SOA or something else, it's secondary to 'are you communicating with the business.'"
As 2010 begins, SOA is more deeply embedded in IT than ever. SOA seems to have survived the near death experience 2009 presented. Richard Soley, chairman and CEO of the Object Management Group (OMG) said anybody who thinks SOA is 'dead' is missing the point entirely.
Ten years ago everybody pronounced CORBA dead, Soley said, and last time the OMG checked there were about 4 billion copies running. If that's 'dead,' he quipped, any technology would want to be dead. Still, he emphasized, SOA itself is not a technology.
"SOA is an approach to better structuring your business operations, recognizing what your business processes are, precisely defining them, capturing them, storing them and recovering them so that you can optimize them and reuse them," said Soley. "And that's just a good idea. To say that's dead is to say motherhood and apple pie are dead."
But SOA as a product you can buy, may be dead, Soley continued. A general adherence to that view, in fact, may be one of the major take-aways of 2009.