In the rush to get online, many companies have put off integrating their legacy systems. Now that customers and suppliers are demanding access to more data, getting the incompatible systems to communicate with one another has become top priority.

Amidst the shakeout of the dot com startups, brick-and-mortar businesses are finding that decades of gathering data have given them a competitive edge. Unfortunately, much of this data is buried inside legacy software applications and databases. What is so problematic about this is that these legacy systems’ designers did not have the foresight to allow them to communicate, either with each other or with customers’ and suppliers’ systems. The crucial element in tapping into such vast information lies in being able to integrate these legacy systems.

Early on, companies migrating online made their first priority the availability of something on the web. “Eighteen to 24 months ago, everyone was afraid of being Amazoned,” says Gaurav Dhillon, CEO of Informatica, a software company that offers data integration and analysis tools. “The VP of e-business was told over and over: ‘Just get us on the Web.’ What we do once we get there and how we tie it to our other channels was not a high-level objective.” As a result of the rush to get established on the Internet, many companies simply swept the integration problem under the rug.

Before they knew it, new e-businesses soon faced customers and suppliers clamoring for real-time access to inventory, pricing and order status data systems. Compounding the added responsibilities of e-commerce, was the risk of getting lost in a labyrinth of incompatible data systems. The previously ignored integration issue had come back to haunt them and was threatening to not only slow down the performance of their applications, but limit future enhancements. These companies took the risk that they could integrate later and miscalculated their approach.

Firms that have already successfully integrated their legacy systems have been those that are willing to step back and rethink their approach. Some, such as Eastman Chemical Co. of Kingsport, TN, have developed comprehensive plans for separating the requirements of their web sites from the limitations of their back-end enterprise resource planning software and database systems. These companies have learned that it pays to build a layer between applications so that data requests go through an intermediate stage. This makes it possible to change and improve the back end without making disruptive changes to the web site. Eastman Chemical has also made strategic investments in an XML provider, which now serves as the connectivity hub to both suppliers and customers.

Complex solutions seem to be the norm. For Silicon Valley Bank, this meant developing a new infrastructure to replace several legacy systems and enacting new guidelines to decrease the chances of problems when discontinuing the use of any particular vendor system. In addition, the bank made Oracle their across-the-board database for any new applications.

In addition to the above approach, there are other recommendations for easing legacy system integration. For the near term, companies should establish and enforce minimum standards for applications, whether they are custom-written or solutions provided by application service providers. They should demand a complete separation of data and logic and also support for a single database management system as well as the capability to import and export using Internet protocols. From within the system, the company needs to be able to select the components for an all encompassing desktop platform that includes hardware, operating system and key application selections.

For e-business applications, the company should construct a layer that will separate new front ends and legacy data sources. A growing market of integration tools has recently been developed claiming to make integration less a matter of programming than of plugging things together. Many vendors are promising integration tools that allow work at a more abstract level when designing business processes instead of having to laboriously map one data field to another. These tools are growing in popularity despite the rise of XML applications because even when the systems to be integrated are XML-enabled, there often remain translation issues. Some of the translation confusion comes as the result of competition between vendors or from consortiums meant to address industry specific needs.

For the long term, companies should build an application architecture that has the goal of complete separation of data, processes and user interfaces. This will enable the easy exchange of applications between vendors and facilitate the development of user interfaces for new media, devices and new applications like instant messaging.

Finally, the company seeking to fully integrate its legacy systems should build or buy next-generation applications that are based on object-oriented design and execution. One of the key advantages of object-oriented software is that its modular nature makes it easier to reuse common logic across applications and across organizations. And, of course, those working on integrating legacy systems need to be able to communicate effectively. In the words of Eastman Chemical’s e-business director, Fred Buehler, “At joint team meetings, the business experts on the e-business side have to give the technical presentations and vice versa. That gives both sides a working knowledge of the challenges faced by the other.” This is the type of coordinated effort full systems integration requires.

Sources: Unleashing the Ancient App
Larry Downes
The Industry Standard, Dec. 18, 2000
http://www.thestandard.com/article/display/0,1151,20615,00.html