News

IBM’s SOA Evangelist Touts Ad Hoc App Dev Vision

• By Stephen Swoyer
• September 19, 2005

Kareem Yusuf, director of SOA product management with Big Blue, says he understands why developers, more than any other enterprise constituency, are skeptical about SOAs, composite applications and loosely coupled application architectures as a whole. It’s a healthy skepticism, he concedes.

“I like to [put] that whole discussion under the heading of governance and processes. SOA is not just about technology adoption or the architectural approach to how you design your systems,” he argues. “It really goes to the core of what are you trying to do as a business and how do you adopt this in a way that is consistent and makes sense and is communicated across the enterprise.”

Nevertheless, Yusuf says, developers should embrace service-enablement when and where it makes sense. And given the premium decision-makers have placed on business flexibility, service-enablement, which in some cases amounts simply to segmenting application interactions into specific internal "services" with well-defined interfaces, makes sense in the overwhelming majority of business cases. “Business flexibility requires flexible IT. The link has to be there. You have to be looking at the business, what you’re trying to achieve with the business, and use that to drive your adoption of SOA by virtue of what your business goals are,” he argues. “This can’t be about a standalone architecture in the sky.”

Yusuf has long been a proponent of composite applications (ganglia of integrated and interrelated services that collectively perform a task, such as a loan-approval process). He declines to speculate, however, about the pervasiveness of composite applications, which, he concedes, must wait on mainstream acceptance of SOAs.

“I think penetration varies. Definitely the early adopters, thought leaders, have been heading down this path for a while. But I don’t know that I can say a lot of organizations are ‘far along’ on this,” he comments. “Most composite applications that I think people are building right now are fundamentally connected with a business process. So you’re trying to do loan origination, or order fulfillment. Typically these business processes involve the interaction of a number of different services to fulfill the end goal.”

Yusuf waxes volubly about a more sophisticated refinement of the composite apps vision, which he describes as “ad hoc” composition. If the yoking together of exposed services into virtual applications comprises the first movement of the post-SOA landscape, ad hoc composition could characterize the second. “[Ad hoc composition is] the other real emerging space, so when you think about a portal being, if you like, a pre-defined composite application, you can surface a number of composite applications in a particular portal and orchestrate those [surfaced composite apps] on an ad hoc basis.”

In the vision of ad hoc composition outlined by Yusuf and other visionaries, Jane C. AverageBusinessUser will have some leeway to construct composite applications of her own, perhaps to suit new or emerging business processes or one-off business tasks.

To some extent, Yusuf argues, ad hoc composition is possible today. He points to IBM’s WorkPlace Collaboration Services, which are based on the WebSphere Portal. “It builds on top of the Websphere portal and really brings to the fore this idea of ad hoc composition,” he argues. “Most of the examples [of ad hoc composition] really sit there in the realm of collaboration. Document publishing, document approval. Collaborative activities like that.”