So here we are in 2004, and what has changed? Well, according to any of the myriad “Best of 2003” or “2003 in Review” columns, quite a bit. In the world of integration, though, wed have to say nowhere near as much as wed like. Oh sure, Web services has emerged as a viable option for many lightweight and distributed integration tasks, integration servers are easing some of the burden of integration at the application server level, and application server vendors like BEA, IBM, Oracle and Sun have all become far more integration focused. But the fact remains that integration is still really hard. Its still time consuming, and the people doing the work are expensive. Software prices are falling, but complexity remains.
To be fair, this is not all the fault of software vendors RedMonk believes that integration in distributed computing will always be a tough challenge. But this hard? Even supposed integration remedies like Web services and Java-based integration standards have seemingly muddied up the waters. So what is an enterprise to do; besides bemoan its fate?
Our best recommendation is to be creative. How? Consider alternative approaches. A couple weve seen recently have some interesting implications for integration:
- Reference Architectures: the definition of a reference architecture usually depends heavily on whos doing the defining. Were generally skeptical of the term, because our experience has taught us that most Reference Architectures are a set of thick three-ring binders that sit on a shelf gathering dust. A recent visit to the Sun iForce Solution Center at the WGBH (a public television and radio corporation) campus in Boston, however, did quite a bit to quell our innate skepticism. What did we see? A working system from the digital video camera through broadcast comprised of hardware and software from partners as varied as Apple, Artesia, Sony, Sun, and Virage in an integrated, functioning system. In other words, this is an architecture that all the partners have certified their products to, and that DAM customers can see in action before they buy. Its a customer reference abstracted, documented, and taken to the next level. It may not be a magical solution to all integration ills, but a customer knows a few things: a.) that the solution can work, b.) that theyve seen it work, and c.) that someone has already gone through the pain of making it all work together. And thats – in our view anyway – worth quite a bit.
Sun is also using the reference architecture approach in its work around RFID tags and auto-ID technology and standards. Thus, for example, it has set up a center in Dallas to help Wal-Mart suppliers with their RFID strategies, and integration with Wal-Mart and the Electronic Product Code (EPC) standards. Wal-Mart is mandating that all of its suppliers use RFID technology to improve supply chain links with the retail leviathan by next year. Sun claims suppliers can test their RFID solution first with Sun and be guaranteed 100 percent compliance with Wal-Mart’s RFID specifications. “We’ve been working with customers to determine the most cost effective way to help with compliance to Wal-Mart requirements and needed to provide hand-on access to the technology and systems needed,” said Sun Auto-ID Business Unit director Julie Sarbacker. “With this new center, we aim to reduce the time and expense suppliers will have to undergo to support Wal-Mart’s requirements.” Sun is also building a similar center in Scotland, UK. - Managed (Hosted) Services: todays attitudes towards managed services are a far cry from the days of Application Service Providers (ASPs) trying to ply services with messages that fell on deaf ears. Managed service providers have benefited from both an economic climate that places a premium on cost efficiency and a gradual abatement of concerns about externally hosted applications. For many of todays enterprise tasks, from regulatory compliance to customer relationship management, managed services are seen as advantageous. Rarely, however, are they seen as beneficial to integration initiatives. We think this is a mistake, however. When was the last time, for example, anyone wondered how Yahoo Mail was integrated with the My Yahoo portal? Probably never, because thats Yahoos problem. Few enterprise-focused managed service providers have focused on this network-based integration advantage yet, so far being content to deliver siloed functionality for the most part, but the opportunity is there. We see baby steps like IBMs Siebel On Demand which is entirely compatible with its enterprise cousin. But we also see enterprise class integration switch providers like Grand Central Communications stepping up to provide managed integration, which is likely to be a more and more viable option going forward. The potential benefit with managed services that integration becomes someone elses problem may yet win the model more friends down the road. GEs Global Exchange Services will certainly be a dominant player here.
Without question both reference architecture and managed services approaches have their cons along with their pros; these are not appropriate for every integration need. The context of each integration project must be weighed; if heavy and complex customization is required, for example, managed services might not be appropriate. Likewise reference architectures are only as good as the products that they certify. Both approaches, however, can alleviate traditional integration headaches, and for many organizations that will be enough. Perhaps more importantly, both seem to be atypical approaches to a long running problem. So while they may seem a little bit too creative, we encourage enterprises to take a long look at these approaches. All youve got to lose is that integration headache. The key insight is that integration is something to think about before, rather than after, you acquire technology. In 2004 and beyond the best way to overcome integration overheads could be to let someone else do the work.
