We're grown-ups now. We don't quote SQL code or babble in remote procedure call syntax. We talk loftily about information integration throughout the business ecosystem. Sure, there's the occasional Freudian slip at the end of a long day of mapping ETL procedures or sorting out XML dialects. But the IT community knows it must dance quickly and easily over such lapses — utterances that a high-tech psychiatrist might ascribe to one's primitive Cobolian subconscious — lest business users become troubled about what ghosts lie beyond the point and click.
To make good on promises to realign IT with strategic business goals, custom-programmed, "spaghetti code" integration must recede into the past. Soon to follow in our distributed, service-oriented world will likely be any programs that tightly couple the getting of information with how it is used or analyzed. Today's vogue solutions, such as master data management (MDM) and enterprise service bus (ESB), address a higher, more abstract level of integration. Peer too long into their confusing pasts and you can easily lose your way. Instead, it's better to focus on how they achieve something new by consolidating earlier strains of technology — each of which has its legacy of successes, failures and incompletes — with more experimental approaches such as Web services.
That these new integration concepts represent technology amalgams reminds us of the larger paradigm shift. The success or failure with early integration routines depended on the skill of individual programmers or developers; MDM, ESB and other high-level approaches rely heavily on shared work. Open standards and even open-source technology (such as Apache or Eclipse) are essential to higher forms of integration. As time goes by, organizations dependent on a small pool of crack specialists for critical integration begin to fret about their vulnerability and increasing costs. Some offshore their legacy coding, but most look at how commoditization and automation can relieve their worries.
One of the biggest challenges in making service-oriented architecture (SOA) work is to avoid giving in to the primitive urge to merely create more processed forms of spaghetti. Mike Ferguson writes in our cover story that many organizations already have several sources of master data, which now require a "shared business vocabulary" to establish one truth for the enterprise. MDM's goal is to improve data sharing across systems in a way that understands the active flow of data, not just its static storage. There's no way organizations can keep information in motion if they must customize each data transformation.
Established vendors are working feverishly to avoid getting stuck in the older paradigm. Some will simply say "me, too" when the trendy acronyms come up, requiring careful examination of such offerings. Others, however, are using their maturity to introduce products that answer more of the questions that arise when implementing new technology. This could be the case with iWay Software, which in February launched a new, "lightweight" ESB that connects to iWay's many existing adapters but is not tightly linked to those or any application server or messaging system. Tight integration is necessary in some scenarios to work with legacy systems, but SOA gurus warn architects not to lose sight of the shared, standards-based infrastructure that supports information flow and plug-and-play application integration.
As businesses develop new products and services, they will undoubtedly reach for unique ways of sharing and using information. IT's job is to make sure that when business users unleash their creative "inner child," the result is not just another plate of cold spaghetti.
David Stodder is the Editorial Director and Editor in Chief of Intelligent Enterprise. Write to him at [email protected].