Gartner: 10 reasons why both sides of the SOA debate have it wrong

Is SOA the greatest thing since sliced bread, or just plain toast? Gartner's Yefim Natis explains why both sides have it wrong.
Written by Joe McKendrick, Contributing Writer

Pro-SOA view: SOA is the greatest thing since sliced bread.

Anti-SOA view: SOA is toast.

SOA moderate view: Let's just worry about baking service orientation into our business processes where we can.

I just had the pleasure of hosting a Webcast keynote with Gartner's Yefim Natis over at the ebizQ "SOA in Action' event, and Yefim did a great job of popping the myths around SOA -- not only among the naysayers, but among the over-optimistic SOA proponents as well. Instead, Yefim urges a balanced middle course with SOA, with a serious emphasis on what it can do for the business.

Here are the 10 most common myths -- promulgated by both SOA "fanatics" as well as naysayers -- that need to be put to rest (no pun intended):

SOA Fanatic Myth #1 - Services were invented in the IT department and are spreading out to the business. This myth assumes that SOA architects and designers "will be bringing solutions to the business that the business itself couldn't invent," Yefim says. However, he observes, "encapsulated functions have existed in business forever. This is how business is structured." Instead, SOA is about improving the "ability of software designers and software architects to model the real world better. Software is not bringing the solution to the business, its better understanding the business."

SOA Fanatic Myth #2 - SOA applications are assembled from pre-built components. "SOA is not a Lego game," Yefim says. "Although service oriented systems indeed include encapsulated components, or services, they also include clients, batch components which are not service oriented, and include legacy systems that need to be connected to."

SOA Fanatic Myth #3 - Sharing or reusing application logic is the main benefit of SOA. "In reality, a successful environment will have reuse of about 30%, so that is a ballpark number where you should feel good about your level of reuse," Yefim says. "If that's the case, it means many organizations will have less than 30% -- so reuse is not the primary benefit, although it is one of the benefits of service oriented architecture. There are many other things, such a making your internal architecture more manageable, having greater extensibility, and applications that function a lot better when they are service oriented."

SOA Fanatic Myth #4 - SOA eliminates the need for application integration. No matter how effective your SOA infrastructure, you're still going to need enterprise application integration, Yefim says. What SOA does do is "introduce a consistency to the architecture, as well as tools and standards that help application integration."

SOA Fanatic Myth #5 - SOA reduces the cost of IT. It may help reduce IT costs in the log run, but early on, "investment in SOA costs in fact costs more," Yefim says. "Not because SOA is more complex, but just because when you do something new, you have to understand new approach, you have to train people, you have to buy new tools -- and that all is costs." What SOA does do is "shift the costs, distribute the costs differently."

Yefim also took the occasion to refute some of the negative things also being said about SOA as well. Here the top five naysayer myths about SOA:

SOA Naysayer Myth #1 -- SOA introduces new complications and new problems. "That might be true, depending on what you were doing before," Yefim says. "After all, complications and problems are all relative to prior experience."  However, he points out, "most issues that have to do with deploying and establishing service-oriented systems are not issues of SOA; they're issues of distributed computing, or of modern grid based computing networks."  Without SOA, he says, companies would "probably be facing the same complications and issues." At least SOA provides a more consistent approach to tackling these problems.

SOA Naysayer Myth #2 -- SOA is nothing new, it's hype, it's taking old wine and trying to sell it in a new bottle. SOA is merely a set of coarse-grained remote procedure calls (RPCs). SOA builds upon earlier models of distributed computing and RPCs, but it's something different, Yefim points out. "SOA is intended to address a business topology of the business functionality of the application, whereas RPCs  were intended to simply distribute an application."

SOA Naysayer Myth #3 -- SOA is doomed because Web services don't work well enough. This widely held misconception is based on the view that SOA is entirely based on SOAP. "There's nothing in common between the two, yet people confuse SOA with SOAP. SOA is not about Web services -- Web services is one of the ways of establishing connectivity between the clients and the services of SOA."

SOA Naysayer Myth #4 -- SOA is hard to sell because the business can't see the benefits. This is probably true for basic-level SOA, but as more companies move into advanced SOA, business benefits will become more apparent, Yefim says. "After all, SOA is an architecture, and the business sees software as a means to a goal, rather than the goal in itself.' However, as SOA begins to support new initiatives such as event-driven processing, business awareness may grow. "Event-driven SOA has very important components to it that allow direct benefits, clear benefits to business operations, to any business that wants to gain control over its overall IT information environment or wants to build situation awareness." Event-driven SOA, Yefim adds, "is the foundation for business activity monitoring, business intelligence, situational awareness. All of these directly serve business."

SOA Naysayer Myth #5 -- SOA is obsolete, and its time to move on. Indeed, the industry is probably ready for a new round of buzzwords, Yefim says. "There's no intrigue anymore in basic SOA.  We know how to do it, it's not talked about as much as before." But, he asks, "What are you going to move on to? The only alternatives you're going to find to SOA are going to be advanced forms of SOA." [See SOA Nay-sayer Myth #4, above...]

Editorial standards