View from the Summit
Architects address critical enterprise challenges at EAS and beyond
by Steve Gillmor
The role of enterprise architect is of equal parts science, management, and art. The science is akin to chemistry—combining ingredients to produce a controlled reaction. The management is about people—mixing minds, experience, and intuition in search of solutions. And the art? That's the secret ingredient that separates the winners from the losers, the magic from illusion.
In music, it's Miles Davis. In film, it's Kubrick or Hitchcock. In baseball, it's DiMaggio or Durocher. For each, the whole was greater than the sum of its parts. By turns seductive, aggressive, persuasive, and just plain lucky, they realized a vision that only seems obvious in hindsight.
An enterprise architecture has many moving parts: heterogeneous legacy applications and infrastructure, innovative (and sometimes disruptive) technologies, the dynamics of open standards and vendor lock-in, and the politics of executive buy-in. No one aspect is more important than another. Cashing in on the promise of services-oriented architecture (SOA) to lower integration costs will mean nothing if the check bounces.
These were the realities confronting enterprise architects from across the spectrum as they gathered for Enterprise Architect Summit 2003, held October 12-14 in Palm Springs, California. From the discussions leading up to the event, a series of overarching issues emerged that traverse the competing constituencies that vie for your strategic attention. To succeed in their organizations, and for their organizations to prosper in the competitive marketplace, enterprise architects will be called upon to address these six critical challenges.
Calibrate Today's Technologies with Tomorrow's Goals
Designing an enterprise architecture is like shooting an arrow at a moving target. You're not blindfolded, but frequently you can't make out the location, let alone the bull's-eye, of the target you're aiming at. It's as though the target was on a fast-moving freight train just pulling out of a distant station, and set to arrive in front of you at noon six months from now.
In this day of reduced budgets and required ROI, it can be difficult to thread the needle between near-term cost reduction and long-range strategic advantage. Only the most targeted architectural initiatives can produce results in a short time window. Vendors are slowing their upgrade cycles, as customers protest the overly aggressive march of new technologies across the enterprise landscape.
But standing pat has its own dangers, some potentially fatal, and not just for the CIO but to the business itself. Take Web services, for example, a technology now coming in for some analytical second-guessing over its viability. Yet a parade of Fortune 500 companies—from General Motors to Merrill Lynch—have slashed integration costs with Web services, while companies such as Amazon, eBay, and Google have used the technology to reinvent or reinvigorate their business models.
Web services may still be a tactical calculation—not if but when to make the move. Those early adopters have taken advantage of substantial standing integration costs to justify the initial XML investment, but others are waiting for tool and platform vendors to ease the development burden. As the leading databases begin to ship with native XML capabilities, ahead-of-the-curve companies such as Salesforce.com are also moving rapidly to adopt true Web services infrastructure.
Another major gating factor is the speed (or lack of it) of standards adoption. Microsoft and IBM have used aggressive market force strategies to tilt the balance in favor of their Web services stack extensions. Sun tries to bake everything into Java through the Java Community Process quasi-standards body, while jousting with open source renegade JBoss over J2EE.
The result: architects have to place subtle bets about the stability of the higher altitude reaches of the Web services stack, those providing secure transactional services that can be quickly migrated along the supply chain. That's where the ROI numbers start to move out of the red and into the black, solving the buy-in problem and financing continued investment.
Balance Legacy and Innovation
Some innovations are easy to integrate. Wireless technologies immediately leverage existing data stores, particularly for mobile and telecommuting employees. But the equation gains complexity when real-time collaboration gains traction. Executive information systems can quickly take over significant developer resources as business intelligence, search, and wireless security upgrades become necessary.
E-mail has become a substantial legacy environment post-Y2K, with a general partitioning of the enterprise marketplace between Microsoft's Exchange and IBM/Lotus' Notes. But innovation has proceeded in fits and starts. The two vendors spent years and millions putting forth URL-addressable models for access to unstructured data (the Web Storage System and Domino), only to abandon those efforts as Microsoft's .Net and IBM's Java/Linux strategies depreciated the earlier efforts.
Tool vendors are a key bellwether as they struggle to migrate their developer communities to Net-aware architectures. Though most trade press and analysts focused on Microsoft's fumbled transition of its Visual Basic 6 community to VB.Net, they missed Redmond's careful positioning of C# as a migration path for Java developers, key packaged apps in the coveted vertical space, and even the open source crowd through Zimian's (now Novell's) Mono.
Sun's VB clone Rave and BEA's Weblogic Workshop are counterattacking by courting a new generation of what might be called soft coders. The strategy is to separate development into at least three areas: the interface, the application engine, and the business rules. By abstracting the complexity of the underlying J2EE infrastructure and its connector-based access to legacy systems, Rave and Workshop hope to empower lower-cost business analysts to speed time to market through modeling, code automation, and rigorous testing methodologies.
Transform IT from Cost Center to Revenue Stream
Successful migration to a Net-based architecture is just the first step in transforming your enterprise. Seeding the XML investment may be the most difficult thing to justify in these bottom-scraping days of the post-Bubble tech depression, but it is one of the most compelling drivers in IT's move to the revenue side. Substantial potential savings in integration costs will be squandered if IT doesn't reinvest in extending XML access out across corporate boundaries. Each new end point—be it supplier, partner, or customer—amortizes the original investment, and soon converts cost into revenue.
Back to top
|