Easing Cobol into a Web-shaped world

Integrating legacy systems doesn't have to be as expensive or painful as it is sometimes presented

Integration -- the connecting of applications, and custom-built software in order to support new or existing business processes -- is doable in any number of ways: just take a look through some of the stories in the Integration channel.

This is probably going to be of particular comfort to the large class of companies that still have (whisper it) Cobol at the heart of their operations. Yes, despite the combined best efforts of both the Unix and Microsoft communities these past 20 years, the COmmon Business Oriented Language is still (at least in legacy form) still with us, and even more surprisingly, providing value for many a system yet. And you thought we'd exterminated all this stuff with Y2K.

Joking aside, the vast amount of heritage system functionality in the world means that there has to be a way to leverage it to give new applications connections to the back end. IDC acknowledges this in a December 2003 white paper on various approaches to integration: "Most organisations believe the primary impetus for integration is to be able to adapt to changing business requirements. Thus we've matured from the concept of integration as pulling in information to map from one location to another, to the concept of addressing processes and change via integrated systems. This requires an organisation to think of its integration solution as a flexible entity, and preferably not so tightly coupled to discrete applications. "

This is the environment non-EAI (enterprise application integration) specialists like WRQ and Micro Focus are working in. They say that the secret to integration isn't necessarily to buy a dedicated integration platform: instead, they support techniques like wrapping and componentisation, which allow Cobol system functionality to get reborn into a more open and cross-application oriented world.

Micro Focus is a survivor of a few turns round the track, and now an independent private VC-backed firm of some 400+ employees. UK director of product strategy, Mike Gilbert, says that his customers face two choices in integration: buy a platform from a company like Tibco or Vitria to act as a central hub for all apps, or use his company's products to bridge mainframe software to the new world, an approach he says Gartner sums up as 'programmatic integration'.

This is fine, he says -- as far as it goes. "It's low risk and cost, but you can run into problems with the security and scalability of the bridging technology. There can be a particular snag if the integration element happens to be associated with a lot of transactions." Many customers, he says, are now looking at migration of the older application, or at least parts of it, to a new platform.

But don't think simple downsizing. That was a huge trend in the 1990s, when companies started baulking in large numbers at the, well, large numbers in dollars they were being charged for mainframe cycles. A lot of apps did get moved, more often rewritten, onto midrange boxes, but at the time PCs just weren't able to take the load of the chunkier pieces of system up at the host level.

Micro Focus customers are solving their integration issues by just this step now, though. International publishing giant Bertelsmann runs an internal IT department it calls the ICS Competence Centre in Vienna. It migrated a raft of existing Cobol apps off its mainframes on to PC servers, a move its CIO Guenther Boedner claims is saving the company around 50,000 euros a month.

The app in question manages central business processes and customer records for some 3.5 million Bertelsmann book club customers of book clubs in Austria, Switzerland, Italy, French-speaking Canada and Poland. It runs to one million lines of Cobol, and was finished a mere 20 years ago. "We expect to achieve an outstanding ROI," he says, and sees integration of these refreshed systems with other applications as "the next step, much easier for us to do now than with any Big Bang approach".

WRQ is a Seattle-headquartered US software outfit active since 1981, whose European HQ is in the Netherlands, and which claims BT, AT&T and Freightliner as customers. Its director of integration strategy, Ron Grevink, explains its approach to integration: "The business cycle has decreased in such a dramatic fashion -- from two or three years to more like six months today -- that IT infrastructure hasn't kept pace. We think Web services is going to be the answer, the silver bullet, but at least so far we have more hype than substance there. We need other technologies to solve the integration problem while we're on the way to the service-oriented future."

Grevink and other Cobol-integration fans like the idea of software as a service just fine -- it's just that they don't see Web services as offering anything more than souped-up interface technology, or in his words "what CORBA or COM did".

The WRQ approach is exemplified in a recent engagement with UK-Canadian tourism specialist Airline Seat Company, which offers services under the 'Canadian Affair' brand. It used integration technology from WRQ to deliver a new e-commerce application that links to essential back-end data and reservation systems. Its UK IT manager Thibault Baradat-Bujoli notes that: "Our problem was the ease of accessing the availabilities of our charter and scheduled flights stored in our back-office proprietary database through the customer Web site."

Two solutions presented themselves: "Either buy the set of APIs from the provider of our back office system which would allow us to do part of this task, or buy a solution which can emulate an operator. We finally choose an integration solution because it was secure and we didn't want to change our legacy systems."

Using the WRQ Verastream integration approach, the company can now link its terminal-based host applications in such a way they can dynamically exchange information with the Web, ERP or CRM applications. Other advantages of this way of doing integration have been ease of changing the Web site's functionality, Baradat-Bujoli adds.

The takeaway thought is that integration doesn’t have to be as expensive or painful as is sometimes presented -- and that there are at least a few technologies that seem to be able to help map the successful old onto the exciting Web-shaped new.