If the move to services architectures is going to shake up and redefine enterprise applications, what will be the end result? JP Rangaswami, CIO at top global investment bank Dresdner Kleinwort Wasserstein (DrKW) gave his prediction last week at a private gathering in London of some of the world's leading enterprise software practitioners. The man who ZDNet's Dan Farber has called "one of the smartest people in IT" is way ahead of most of his peers in embracing current changes in enterprise IT, most notably the use of open source software, as I've discussed in another write-up of his talk posted on my Loosely Coupled blog.
But what I found most interesting in the context of what I've been writing about here on ZDNet were his views on the future shape of enterprise applications. Whether you call them Web 3.0, enterprise 3.0 or something else, there are some powerful forces in motion that are going to fundamentally shake up our current view of enterprise applications, replacing the current demarcations — ERP, CRM, BI, collaboration, and so on — with something new. He identified four core families of requirements that he believes will be left when all this shakes out, and which are already shaping DrKW's internal architecture:
Publishing: Any application that generates data will act as though it's a content publisher, using RSS or similar to publish its data. The significance of this is that it reduces all of these applications to the level of raw feed generators: "You can't differentiate, it's just content." This puts these applications down in the commodity API services layer I described in my posting last week on What to expect from Web 3.0, which I suspect is going to be bad news for any vendors that rely on this layer for software licence revenues.
Discovery: This is the application that gives everyone a "Google experience" — a single, homogenous database where everything is stored and where everything is discoverable. Rangaswami noted that how you implement security can easily get in the way of this objective. Make it too much of a fortress and the risk is "we put the data in that worst nightmare of walled gardens: ours." Therefore, the starting principle has to be that the information database is open access, with access and authorization controls added only as necessary for specific items or classes of information.
Fulfilment: This is the application that makes things happen, most notably for customers. Identity management plays a crucial role here, controlling the catalog choices that are shown to each user and their rights to approve shipment.
Conversation: All the channels of collaboration between people, either inside the organization or beyond its walls.
"These four application families are the vision that we want to get to," he concluded. How long will it take to get there? "I'm calling it four years. It could be ten."
His talk was the keynote of an invitation-only summit held at DrKW's London offices last week, which has begun work to draw up a practitioner-driven manifesto of key principles for enterprise application development. The group will be publishing a series of documents over the coming months, setting out definitions, common fallacies, design principles and a manifesto for systems architects setting out to build the next generation of enterprise IT.
What do you think? Do you agree with Rangaswami's four pillars? Are they the right pillars, and the right number? How do you think enterprise IT will have changed in four or ten years' time?