Content Integration: Are You Ready for It? Is It Ready for You?

Page 2 of 3

Appservers: The Corporate Mixing Bowl
Application Server (appserver) vendors like BEA and IBM have a somewhat more straightforward story, targeted squarely at architects and developers. Since your appserver increasingly serves as the foundation for mixing all your other applications—and now comes with robust portal features too—shouldn't it also serve as the basis for integrating all your content applications as well? And it is the appserver vendors—along with Microsoft's .NET initiative—who have been the most aggressive in promoting Web Services as an integration paradigm.

This story is attractive on its face, but it doesn't take too many chapters to realize that for content integration it ultimately means doing an awful lot of custom coding. Sometimes custom code is essential, but if software exists to do the same thing, do you really want to reinvent the wheel? Moreover, these vendors don't always have a strong answer for connecting to non-Java-, non-.NET-, and non-SOAP-enabled systems. Appservers uniformly offer powerful EAI tools for integrating transactions, but as we'll see shortly, they are just now getting into the CI game, typically via partnerships.

Vendor Suites: ECM in a Box?
Some content management vendors have a different recommendation for content integration: "just buy it all from us." In an attempt to become "enterprise" CMS providers, such companies as Documentum and FileNET have developed integrated systems—either through partnerships, or, increasingly, acquisitions—that address most or all of the content lifecycle from authoring/scanning to syndication and archiving. For companies owning existing systems from other vendors, though, the prospects of outright application replacement are often less than appealing, and savvy CIOs know that product suites assembled via acquisition may not really be as well integrated as they appear.

Other CMS vendors take a slightly more modest approach. Rather than underpin your entire content lifecycle, they develop a series of connectors to other applications so that your CMS becomes a kind of content hub. For example, Day Software has developed a "content bus," to draw content from other appli- cations. Vignette, Tridion, and Stellent offer a variety of "adapters" for bi-directional content flow among various systems, including ERP installations. Note, however, that these connectors don't allow different applications to talk to each other—just to your CMS.

The CI Alternative
A new breed of software company is emerging from former professional services firms that tried to tackle these difficult integration issues in the late '90s. Their CI offerings don't replace existing software installations, but rather, ride on top of them to provide a more integrated set of views and services without undermining investments already made in key point solutions. This new crop of CI vendors remains quite small, but they are slowly feeling out the architectural approaches and implementation lessons that could drive substantial future activity when the IT sector emerges from recession.

Several companies now call themselves CI providers; later I'll contrast the approaches of two of them, ContextMedia and Venetica, as representative of the field at large. Other significant players include Agari, which focuses on cross-media tools, and Ascential, which comes from a data extraction and cleaning background.

ContextMedia: Think Category, not Repository
Founded in 1999, ContextMedia is rather unique inasmuch as it offers both a server product, Interchange, and optionally, an add-on client application called InterShare. Both do pretty much the same thing: provide a "unified view" of the content residing in disparate repositories. You can use this view to drill down to find resource, then check the content out of one repository and check it into another—or simply copy it, depending on your rules. Interchange can also "package" up sets of related content, ship them somewhere, and track how they get used.

At this point, you're probably suspecting that the whole thing must run on metadata. Well, you're right. Interchange builds and maintains a major central metadata store (for all the assets among the repositories it connects to) that allows you to search, view, group, and manage remote content unobtrusively. Capturing all that metadata—and normalizing it around the Dublin Core or some other universal schema—is the first critical task of any ContextMedia implementation. Not surprisingly, the company partners closely with auto-classification vendors.

Having done this, Interchange enables you to look at your content in a new way: according to category, rather than repository. Now we have a possible solution for ACME's customer service manager—her new hires can browse corporate assets by topic.

In addition to pre-built adapters to Interwoven, Artesia, and similar packages, ContextMedia provides a framework for writing custom connectors to legacy systems. In fact, more likely than not, it is ACME's defunct CRM system that will impel the company towards a CI solution. According to Jonathan Brown, director of product strategy at ContextMedia, "what really tips a company to content integration is an unsupported legacy system—the company has already invested in workflow and user screens, and they don't want to give that up."

But what about the ACME marketing staff who want to publish images from Artesia on the public Web site? Their choices illustrate some of the trickier areas around business rules and asset management.

Normally, the marketer would log into Interchange or Intershare, search and retrieve the image from Artesia and then… well, then what? They could copy the image to the Vignette asset library, but now there are two versions of the file, although Interchange keeps a reference between them. Alternatively, the ACME staffperson could just set a pointer to the image and instruct Vignette to retrieve and serve it up at deployment time. But what if the source image changes before then? Will that matter?

There are several other possible permutations here, some of them involving workflow triggers (which could be either highly useful or highly mettlesome), but they all center around the trade-offs between usability and immediacy on the one hand, and maintaining authoritative versions and referential integrity on the other.

Specialists at ContextMedia suggest that most companies seem willing to exchange the creation of multiple instances of an asset for the flexibility of using it seamlessly in diverse environments. Most organizations "only need to copy files 10% of the time," notes Brown. Nevertheless, without actually moving content around, a CI deployment risks turning into nothing more than a fancy search engine.

Using ContextMedia, the choices could fall to the end-user at their desk. Although you can configure the product to work within an application so that Vignette could "see" the Artesia and Stellent repositories, Intershare is designed to function as a standalone product—available as a separate application to the beleaguered ACME employee simply trying to find and use content to complete a task.

This notion of a separate "master interface" for CMS might be ideal for some firms, but unrealistic for others. For IT groups, it presents yet one more user-facing application to support—possibly with a separate client installation. For end-users, Intershare poses another system to learn. Nevertheless, ContextMedia is gaining traction, especially among media and publishing companies, so it remains a solution to watch.

Page 2 of 3