Two weeks ago I was in Phuket Thailand, sharing some research about SOA adoption in Asia. Our team had recently conducted a survey of over 300 A/P IT executives, and one of the questions we asked was: “How important is it for your IT organization and IT suppliers to adopt a service-oriented architecture?”. An overwhelming 84% responded either “important” or “very important”. This follows our U.S.-based survey earlier this year, in which over 75% responded likewise, and we have seen similar as I wrote earlier this year, SOA has clearly emerged as the IT blueprint for CIOs, particularly in large enterprises – chiefly because of its promises of 1) simpler/faster application integration, and 2) simpler alignment and linkage of IT resources in support of business processes.results in Europe. This supports our view that,
But, in spite of CIOs’ avowed, overwhelming endorsement of SOA as an architectural principle, SOA adoption has been painfully slow. We estimate that 10% or less of shops have implemented SOA in any broad fashion. The obvious, critical questions are: Why is user adoption of SOA so slow, and how SOA will finally achieve widespread adoption in the marketplace? Here’s what I see, now and for the future of SOA adoption:
- SOA Adoption Today Is Most Essential for – and Most Adoptable by – Software Vendors Themselves (vs. Mainstream Users). Today, taking advantage of SOA’s promises of faster integration and deployment, and lower maintenance costs, is much more important for IT industry players themselves than their customers. This is for two obvious reasons. First, software vendors – whose core business depends on their ability to quickly create, and steadily improve, software – spend a fortune living with the cost, speed and complexity problems SOA is aimed to solve; and so they have the greatest business motivation to adopt this new approach. Second, implementation of SOA today is still complicated, depending on the use of servers, frameworks and toolkits, that require significant developer skills – something software vendors have in good supply, unlike most user IT shops (see next point). And so we see the current stage of SOA adoption being essentially industrial in nature – focused on ISVs, using the technology themselves to develop the next generation of SOA-based commercial software offerings.
- Mainstream Users Will Adopt SOA Broadly When It Is Delivered “Under the Covers”, In Packaged Solutions. For the past 20 years, the software industry has successfully convinced most customers to get rid of most or all of their own developers, and – instead – leverage packaged applications. This strategy has certainly been a sensible choice for most customers, and a profitable route for vendors. As a result, though, the industry is now looking at a customer base that is largely incapable of adopting SOA – or really any software technology – until it is delivered “under the covers”, within next-generation packaged solutions. Technologies that are available as “raw materials” – primarily in the form of platforms, frameworks, and toolkits – that allow customers to “roll their own”, are, by definition, limited to adoption by customers who are ISV-like, or are willing to pay SIs a lot of money to “roll it” for them. Those customers are in a very small – albeit economically important – minority (IBM, for one, has had great success in penetrating this up-scale customer segment with its WebSphere products). Widespread SOA adoption with really only begin when the ISV community itself delivers SOA “under the covers”, that is within next generation packaged solutions and, in effect, largely invisible to most customers. The good news is that the major SOA platform players- including IBM, SAP, Oracle and Microsoft – are busily developing large partnership ecosystems of service-enabled applications, as well as moving their own applications to a service-oriented architecture – so we’re getting closer to this first wave of mainstream, solution-based adoption.
- The Second Big SOA Adoption Wave Will Be On the ‘Net, via SaaS. Within the past couple of years, a new dynamic has emerged that promises to give the ultimate boost to SOA adoption – the growth of the Software-as-a-Service” (SaaS) model. We believe SOA and SaaS will be increasingly, intimately tied to each other, for two main reasons. First, while packaged solutions (that drive the first wave of mainstream SOA adoption) will make it significantly easier for users to adopt (and fund) SOA, adoption will not be a trivial exercise. On-premise implementations of service-oriented solutions will still require IT shops to upgrade technical skills, as well as invest significant up-front costs for migrating to the new-generation solutions. For many shops, this will be the preferred path – and not much different from earlier migrations to applications based on new architectures (e.g., client/server, web-based). However, the SaaS model will, we believe, greatly accelerate the adoption of SOA-based solutions by reaching users who are more cost- and skills-constrained (including the “SMB Long Tail”), by offloading more of the burden of SOA adoption from users, onto SaaS service providers. In the SaaS model, the service provider hosts the new SOA/services infrastructure, offloading much of the capital outlay and technical skills upgrade burden from the user. The second important reason SOA and SaaS will be tied to one another, and accelerate each other, is that SOA will increasingly be a critical foundation for delivering SaaS. SaaS service providers will primarily compete with each other on the basis of the richness of their solution ecosystems; that is, customers will gravitate toward service providers whose SaaS solution portfolios best match their business needs. SOA will be the most practical way for service providers to support the massive scale of integration that these large solution ecosystems will require.
So, is SOA the next big IT architecture, or is it just the latest hype? It is certainly worthy of debate, since SOA is both wildly popular in concept, but not nearly so popular when it comes to actual customer commitment. In our view, SOA is – in fact – the new foundation for the industry, and the current concept-commitment gap is simply an artifact of an adoption ramp-up in its early stages. In the current, “industrial” stage of adoption, SOA is largely a programming construct, implemented primarily through fairly complex programmer toolkits and servers. We are currently shifting to the second, commercial adoption stage, where SOA will be primarily consumed in the form of packaged applications and solutions. In its broadest adoption stage – which will come online as fast as the SaaS model is adopted and pushed with enthusiasm by ISVs (e.g., SAP, Microsoft) and partners – SOA will be adopted largely in the form of actual business and consumer services.
In short, SOA will really reach its potential, when “services” is less about a programming model, than it is about delivery of real consumer and business services. And, with the steady growth of SOA-based solutions – both on-premise and (especially) online – that day is finally coming.