Will legacy applications ever die? By now, you might think that most companies would have replaced software programs written in such musty languages as COBOL and Fortran. After all, many spent millions during the Y2K hysteria pulling out tangled systems and marshaling armies of programmers to rewrite software. And as businesses came to rely on the Internet, the software that underpins many operations got (and continues to get) a serious overhaul.
Yet the old applications are still with us, to a degree that might surprise you. According to Aberdeen Group, mainframe applications written in COBOL still process 70 percent of the world’s business data. Many companies continue to run programs that are well over 20 years old. Part of the reason is that earlier efforts to replace old systems actually touched only a small fraction of the applications in use. And, hoping to preserve some features of the old software, companies weren’t always disciplined about discarding old systems after putting in new ones.
The presence of older applications becomes even greater when you consider the broader definition of “legacy” that many technology experts advocate — not just homegrown systems, but any packaged application that is more than two releases old or is no longer fully supported by the manufacturer. This definition encompasses even Windows-based products and some newer enterprise systems. And it means that many IT executives have come to feel like gardeners, yanking weeds only to find new varieties sprouting.
The result of this stubborn persistence is that companies continue to pay the cost of maintaining old applications (according to Gartner, between 40 and 60 percent of an IT department’s budget is for this purpose) and struggle to modify them to support changing business needs.
Strength and Flexibility
Unglamorous though it may be, many software companies have been hard at work devising new answers to the problem. In fact, legacy systems have become a fecund area for innovation. The technologies now available offer an appealing proposition: the ability to breathe new life into wheezing applications. In other words, new methods allow companies to take advantage of the strengths of legacy applications — typically their processing power and business-specific customization — while making them more flexible.
“The fact is, legacy applications remain among the organization’s most valuable assets,” says Dan Sholler, a vice president at Meta Group. “They still do what they’re supposed to do. Why spend money to fix what’s not broken?”
To understand the appeal of this new take on legacy systems, consider the problems with removing older applications in favor of packaged software. Known in the IT world as “rip and replace,” this approach can be as jarring as it sounds. Because the applications often contain valuable historic data, putting in a whole new system — which may take years — is not just expensive but also often wreaks havoc with business processes. “It’s like a heart transplant,” says Chris Selland, vice president, sell-side research, at Aberdeen. “I’d probably be a somewhat better athlete with a brand new heart, but it would be extremely disruptive to just stick one in.”
One alternative is to integrate a company’s applications — old and new — at the surface level only. Screen-scraper software, more formally known as programmatic interface technology, lets companies link multiple applications by using just the information available at the screen level, or user interface. In essence, the software acts like a user, logging into legacy applications, entering potentially arcane commands, and returning information to a single interface that shields employees from the behind-the-scenes complexity.
One version of this technology is offered by Jacada, a provider of noninvasive integration software with offices in Atlanta, London, and Israel. The Jacada Fusion product allows companies to integrate not only text-based mainframe applications but also Windows and Web-based programs.
The Hartford, an $18.7 billion insurance provider, is using this technology as part of a larger effort to give its insurance agents easier access to its legacy applications. According to Mike Byam, CIO for business insurance, the company uses Jacada Fusion to provide a user-friendly front end for agents who need to determine whether The Hartford is willing to underwrite a certain risk. Instead of having to log in and out of many different applications, the agent can go to a single screen while the software digs into the legacy systems on his or her behalf.
Another, related way of milking legacy applications is through the use of business rules engines, a technology enabled by Web services. Business rules are the policies and procedures of a company — how it creates a price or defines a product, for instance — that are buried within the code of a legacy application. In the case of a credit-card provider, for example, this might include the specific conditions under which the company will grant a customer a higher credit limit.
With the help of business rules software, a company can turn that feature of the legacy application into a Web service. A Web service is a discrete component of a software program — such as the subroutine that determines a customer’s credit limit — that is standardized so that any other computer system can use it. When a company has created a number of Web services, they can then be rearranged into what are essentially new applications. In effect, this means that disparate legacy applications can now work together in new ways. “You can leverage the asset you’ve already purchased — one that may be 20 years old — and repurpose it for use through a Web services framework,” says Eric Austvold of AMR Research in Boston.
Insurers such as The Hartford and Horizon Casualty Services are embracing this approach. Horizon, a subsidiary of Horizon Healthcare Services Inc., is using software from RulesPower to replace some of its legacy applications. “Making changes to the existing applications is very complicated — change one thing here and many others have to change, too,” says John Oliveira, director of operations. “With the rules engine approach, the same task is simpler and faster.”
Another Layer of Dirt?
There are limits to this new software. For example, because it doesn’t actually touch the legacy applications, screen-scraper technology doesn’t provide true integration. Back-end integration might be desirable for a call-center operation: if an operator enters a change of address, you want to be sure that all relevant applications are updated. In addition, neither screen-scraper technology nor business rules/Web services technology helps a company significantly reduce its systems maintenance costs, since the old systems remain in place.
“You have to look at the business driver for extending your legacy systems,” says Chris Jones, senior vice president of research at Aberdeen. “If there’s data in your systems that you need, then you might just look at Web-enablement of legacy applications or user-interface improvements.”
Furthermore, if new technology isn’t added judiciously, it could complicate matters in the long run. “With new technologies, you have to be careful that you’re not just adding another layer of dirt and creating a mess for the next person,” says Mike Childress, a vice president at Electronic Data Systems.
ValueOptions, a managed behavioral health-care services provider based in Norfolk, Virginia, has addressed these challenges by consolidating legacy applications and then using Jacada’s technology to update the remaining applications. Over six years, the company has consolidated seven different legacy systems to just one, and now uses Jacada Fusion to convert text-based screens of the remaining system into a simpler interface for customer-service representatives.
According to CFO Thomas Oram, he and the ValueOptions executive team considered the rip-and-replace alternative, but rejected it. “We thought there was too much risk in going to a completely new system,” he says. “There are decades of data in our legacy system and customization that would be costly to replicate. It was more attractive to take the best of our legacy systems and add state-of-the-art technology on top.” Still, the project hasn’t been cheap: the company has spent approximately $20 million to consolidate its systems.
Don Durfee is research editor at CFO.