Was 2008 the year in which service-oriented architecture (SOA) began to lose some of its once-brilliant lustre?
One key piece of research certainly gave the impression that what was once heralded as the secret to business-responsive IT – dividing application functionality into discrete, interchangeable services – had been found too ambitious, complex and difficult by at least some of the user community.
The headline finding from analyst company Gartner’s 2008 SOA adoption survey was that the proportion of organisations planning to adopt SOA in the next twelve months had fallen from 54% in 2007 to 24% in 2008.
That chimes with comments made by Eric Guilloteau, CEO of user interface service vendor Corizon. “The momentum behind SOA is not as strong as it was,” he told Information Age in November 2008. “There was a lot of expectation that SOA would be more popular than it is. There was a lot of talk, but adoption is about a year behind.”
The most commonly cited reasons for neither deploying nor planning to deploy an SOA were a lack of internal SOA expertise (60%), no perceived business value (50%) and the prohibitive cost of SOA (44%), Gartner found.
But SOA apologists were swift to point out that this was no proof of its demise. After all, the same report also found that SOA adoption (i.e. the proportion of respondents who are currently using SOA) had risen from 37% to 53%.
If more companies got around to starting their SOA projects this year, they argued, no wonder the number planning to implement in the future has gone down: the figures were merely evidence of SOA’s maturation.
That argument does not quite balance: Gartner also found that the number of organisations with no plans for any SOA projects jumped from 6% in 2007 to 16% in 2008, suggesting that around 10% of organisations have made up their mind in the past twelve months that SOA won’t work for them.
But as even that statistic reveals, SOA is still very much in the frame.
The findings of the Effective IT Survey are more positive for SOA. Last year, 23% of respondents had already adopted the model; this year, that figure was up to 36.3%. Meanwhile, the number of respondents planning to deploy SOA in the next 12 months was up three percentage points to 24%.
Explanation for this discrepancy can be found in the geographical analysis of Gartner’s numbers. Among the European subset of respondents to its survey, the most likely to overlap with the Effective IT Survey’s
A mature movement
So reports of the death of SOA may have been exaggerated. What may in fact be happening, however, is the term itself approaching the end of its useful life.
“More and more, SOA is part of the furniture,” explains Neil Ward-Dutton, an analyst at IT advisory Macehiter Ward-Dutton and SOA expert. “Many organisations making new IT project investments are likely to be working with SOA principles, whether or not they made an explicit decision to go down the SOA route.”
Examples include new SAP or Oracle deployments, which will include components of those application vendors’ respective SOA platforms, whether or not the customer has an explicit SOA strategy.
That effect is magnified by the fact that the number of technologies that could be defined as constituents of SOA is many and varied. SOA has always been a catch-all term that includes many different technological approaches, but it is becoming abundantly clear that there is no consensus on how it should be done.
“Many organisations making new IT project investments are likely to be working with SOA principles”
In Gartner’s research, for example, only 60% of those companies adopting an SOA were doing so using web services, although they are the most popular kind of service. Similarly, the enterprise service bus, argued by some to be an essential component of ‘real’ SOA, was only used by a quarter of SOA adopters.
One palpable trend in Gartner’s numbers was a shift in the preferred development and integration language among SOA adopters over the past three years. In 2006, JAVA and JAVA-related languages were the de facto platform for SOA development project, but since then their proportional use has fallen as Microsoft’s .NET language has grown.
The report’s authors wrote that the underlying trend driving those numbers was not obvious. “It may be that organisations planning to implement SOA choose Microsoft more often,” the report said, “or it may be that those organisations using Microsoft are encouraged by some characteristics of the environment to use SOA.”
More evidence of the demise of SOA as a cogent concept came from the financial performance of SOA-related companies. Two companies that have closely associated themselves with SOA are Tibco and Progress Software, both of which sell the infrastructure machinery required for a textbook SOA.
In October 2008, Tibco revealed that its SOA-related revenues had grown by just 5% year on year, hardly suggesting a ‘paradigm shift’ in enterprise computing. Fortunately for that company, its other lines made up for that softness; Progress Software total revenues were up just 4% in the same quarter.
Of course, there are companies that are making money from SOA infrastructure. Oracle’s middleware division, which now includes acquired SOA infrastructure and tools vendor BEA as well as its own SOA middleware products, grew 35% year on year during its first quarter of the year, ending August 2008.
But that only serves to underline the point that while SOA is certainly prevalent, and increasingly so, its validity as a ‘movement’ above and beyond the product strategies of the IT infrastructure vendors seems less than it once was.
That doesn’t mean the SOA movement is wrong, however. Indeed, developments in web-based computing validate the core arguments for SOA. For example, the logical extrapolation of cloud computing (the enterprise IT ‘movement’ that has usurped SOA’s buzzword status) is arguably an infrastructure composed of multiple constituent services.
“SOA continues to be really important,” says Ward-Dutton. But he acknowledges that, in light of the economic climate, SOA projects may be in for a troubled time ahead unless they are linked to a clear business benefit.
“For those companies that have done a decent job of explaining the business value of SOA to business leaders, it will probably be an even more visible element of IT strategies in 2009 than in 2008,” he explains.
“But for those companies that have pursued SOA from purely an IT perspective, there may be greater challenges in pushing SOA initiatives forward as infrastructure spending gets squeezed and as more spending gets diverted to more clearly business-linked projects,” he warns.