Blogger: Anne Thomas Manes
My pronouncement of SOA's death back in January sparked quite a lively debate. According to Joe McKendrick, the debate is still raging.
Just in case anyone is still confused by what I said/meant when I said "SOA is Dead; Long Live Services": "SOA" as a term has lost its luster, but "SOA" as a practice is essential for all organizations going forward. Many organizations have invested millions into SOA, and they have little benefit to show for it. Some organizations are worse off than when they started. Given the tight economy, business people aren't particularly interested in pouring more money into what looks like a sinking ship. If you want to get funding this year for your SOA initiative, you should probably avoid using the word "SOA" and instead focus your efforts on building "services" that deliver measurable value to the business.
Now, that's not to say that no one has succeeded with SOA. IBM and Software AG both trotted out a number of great success stories this month at their conferences, IMPACT and SOA Summit respectively. In particular, everyone was raving about the Coca-Cola case study at SOA Summit. Burton Group also has some great success stories on deck for Catalyst 2009. But true SOA success is hard to come by. Our definition of "success" is positive return on investment. If you've invested $5 million over 5 years, your initiative is not successful unless you've generated >$5 million in positive business outcomes.
Which brings me to the real topic of this post: Proof that SOA is still dead.
According to a recent Gartner survey, 40% of users do not measure how long it takes to achieve a return on their SOA investment. The survey also shows that 50% of those companies that have not yet started a SOA initiative did so because they could not articulate and demonstrate its business value. Without a means to measure value, SOA initiatives are doomed. Quoting from the IT PRO article summarizing the Gartner survey:
Massimo Pezzini, research vice president and fellow at Gartner, said that many companies were approaching SOA projects with excessive expectations and little awareness of the effort, resources and time needed to achieve any benefits.
“Some SOA projects are perceived to have failed when in fact there are simply no well established metrics to evaluate success,” he said. The pressure of such expectation, coupled with the promises of SOA technology vendors, were leading companies to over-spend on technology but under-spend from an organisational and governance viewpoint, Paolo Malinverno, research vice president at Gartner added.
“So they come to the conclusion that SOA is expensive and doesn’t deliver,” Malinverno said.
As David Linthicum said in his commentary on the report, "Shame on you guys!"
There was a bit of discussion in the Twitterverse yesterday among @madgreek65 @richardveryard @neilwd @davidlithicum @jhurwitz @Lowrain @cobiacomm and myself (@atmanes) on this report. I loved Neil Ward-Dutton's characterization of ROI measurement as a "minority sport". David has written quite a bit on measuring SOA ROI. As he said in one of his tweets, "Google SOA, ROI, and Linthicum". Burton Group subscribers should also take a look at "Building the Business Case for Service Oriented Architecture Investment". I also have a document scheduled to be released next month on developing business value metrics: "Using Metrics Effectively: Proving and Improving the Business Value of IT". We'll also be talking about business value metrics at Catalyst. We have a few good case studies to share.
I've seen two other recent indications of proof that SOA is still dead:
1. Gartner just recently published its annual assessment of the application integration and middleware (AIM) market, which experienced single digit growth in 2008. According to a review of the report by Application Development Trends, "Middleware Market Hits the Brakes in 2009", Gartner is projecting a 0.8 percent decline in the AIM market for this year.
2. Gartner reports that IBM holds 30.8% of the AIM market. Another market study by Report Buyer asserts that IBM holds 70% of the SOA infrastructure market. So IBM sales should be a pretty good indicator of the SOA infrastructure market. And according to a tweet from James Governor at IMPACT, "Robert LeBlanc GM, software sales says clients are buying SOA in smaller chunks now." (I interpret "buying SOA" to mean "buying SOA infrastructure software", because we all know that you can't buy SOA.)
Now, of course, plenty of people continue to refute my claim that SOA is dead. First in line is Steve Mills, Senior VP and Group Executive of IBM Software. Joe McKendrick asked him for his take on the debate while he was at IMPACT, and then faithfully wrote up his response. Steve vehemently supported my claim that SOA as a practice is essential going forward, but he said nothing to refute the claim that business people have "come to the conclusion that SOA is expensive and doesn’t deliver.” Robert LeBlanc's report on SOA infrastructure sales demonstrates that IBM is certainly feeling the effects of business people's disillusionment with SOA.
Other market sizing research firms disagree with the Gartner report. For example, the Report Buyer survey cited above predicts that the SOA infrastructure market will grow at an average rate of 17.1% over the next 6 years. Assuming the economy recovers within the next 2 years, this prediction seems reasonable -- but the growth will almost certainly be back-loaded. I seriously doubt that we will see double digit growth this year or next. I think Gartner has a better view of the more immediate future.
My friends over at Forrester have also refuted my claim--sort of. Randy Heffner published a document last week entitled, "SOA is Far From Dead, But it Should Be Buried." The title is a bit misleading. When Randy says SOA should be buried, he means that it needs to be "buried inside a larger vision". Actually, I think I said that in the original SOA Obituary:
Successful SOA (i.e., application re-architecture) requires disruption to the status quo. SOA is not simply a matter of deploying new technology and building service interfaces to existing applications; it requires redesign of the application portfolio. And it requires a massive shift in the way IT operates. The small select group of organizations that has seen spectacular gains from SOA did so by treating it as an agent of transformation. In each of these success stories, SOA was just one aspect of the transformation effort. And here’s the secret to success: SOA needs to be part of something bigger. If it isn’t, then you need to ask yourself why you’ve been doing it.
The latest shiny new technology will not make things better. Incremental integration projects will not lead to significantly reduced costs and increased agility. If you want spectacular gains, then you need to make a spectacular commitment to change.
Randy does refute my claim that organizations are reducing their SOA investments, and he backs it up with data from a survey of 2,227 IT executives. Joe McKendrick summarized the report here. According to the survey, 24% of users say that SOA has "delivered most or all of the benefits expected", and 36% say it has "delivered enough of what they expected to justify expanding their SOA adoption". And more to the point, only 1% say they have “seen little or no benefit” and are cutting back on SOA efforts. I presume that the remaining 39% have realized modest benefits at best, but expect to maintain current investment levels.
A 24% success rate is a little higher than what we have directly observed, but not horribly out of line. Besides, Burton Group doesn't run statistically relevant surveys. The 1% "cutting back on SOA efforts" is much lower than our observations, though. Many of our clients (Global 2000 companies and government agencies) have reduced their SOA investments this year. It also contradicts the drop in sales reported by IBM and predicted by Gartner.
I'd like to see how the middle groups (75%) in this survey correlate with the 40% of users that aren't measuring ROI. We've found that many organizations can't definitively say how well their SOA initiatives are going because they lack hard metrics and baselines.
It's quite possible that these organizations have reduced spending on SOA infrastructure, and what they mean is that they are applying SOA practices in a larger percentage of projects. That's what John Rymer implied at his keynote speech at SOA Summit. As reported by Joe McKendrick:
In his presentation kicking off the summit, John Rymer said that Forrester’s surveys show plenty of strength in SOA adoption plans — for example, 27% of the largest enterprises currently have SOA in place, and 33% are committed to moving in this direction. SOA principles themselves did not die, but rather, “SOA died a marketing death,” meaning that the approach has become so vital and basic to enterprises and as a part of packaged applications that marketers have moved onto the next big thing. “When a technology becomes vital, it dies in a marketing sense,” he explained. “It’s time for SOA to ‘die’ since it’s not distinguishable anymore since everybody’s using it.”
I certainly hope John's assessment is true.
SOA is still the most popular search term on the Burton Group research site. And we're definitely very busy assisting clients with their SOA initiatives. But we're still seeing a lot more stalled efforts and failures than success stories. So I don't think we're out of the woods yet.