|
INDUSTRY WATCH: Thoroughly Modern Mining
By David Rubinstein
 January 15, 2004 — Object Management Group Inc. has undertaken an ambitious project, called Architecture-Driven Modernization, as a way to make current an organization’s legacy systems—“any production-enabled system regardless of language or platform.” As I discussed in the last column (“Two Sides to Every Transformation,” Jan. 15), Phil Murphy of Giga Information Group claims the idea is too broad and, because the initial effort will not delve down to the code level, “fatally flawed.”
Into the middle of this imbroglio has walked Blue Phoenix Solutions Inc., which recently joined OMG and is a part of the AMD effort, yet which also thinks Murphy has raised some good points in his counterargument. Blue Phoenix has pulled off some recent acquisitions to move itself squarely into the COBOL regeneration process.
“We place ourselves in the modernization space,” said Ted Venema, Blue Phoenix’s senior vice president of product management. “Our general feeling is that screen-scraping is a good first step, but clearly it’s not enough. Step two is automated migration—pump something in and get something else out. This works well when you’re not crossing architectures. COBOL to Java failed miserably because there were real tough architectural issues.”
Among the problems are moving from a CICS world to a GUI, event-driven one; dealing with a procedural language such as COBOL in new architectures that have been written for object orientation; and moving to a process-driven approach from a screen-driven approach, Venema said.
“Redevelopment is the way to go,” he added. “You mine out of what you have as much as you can, the business-flavored code, and leave the technical code behind. Otherwise, you end up in a no-man’s land between two systems. You want to mine the screen definitions, the data definitions, procedural and data flows, the things that are really relevant.”
The OMG effort “is too big,” Venema said, agreeing with Murphy. “The way we approach it is IT discovery. You put all the assets into a relational repository. We’re working with [OMG] to define what the repository should hold.” But Venema believes, as does the OMG modernization task force, that going down to the paragraph level is deep enough. “We just want to know how things hang together, such as which programs call which programs.” From this information a model can be created, Venema explained, and tools for code maintenance or analysis can work off the model. “There just is no clear idea yet what the model should be.”
Ed Gentry, vice president of product management at Blue Phoenix, discussed the efforts of company partner Cook Systems International Inc., which has combined the Blue Phoenix rapid application development and legacy modernization technologies with its own knowledge of systems and applications to offer a code maintenance solution at a greatly reduced rate.
“Maintenance of these COBOL applications is expensive because the app systems are fragile and not flexible,” Gentry said. “What I need is to be able to rearchitect the application, using the existing app as a specification, and create a new application in the new architecture with the same functionality.” Gentry said Cook, using a RAD tool, was able to reassemble one particular COBOL application as HTML.
This type of effort, Venema said, is less ambitious than OMG’s goal to modernize everything, but can be implemented more quickly and accomplish more in a specific system. “We’re not interested in everything in your system,” Venema said. “At least a third of that is throwaway stuff, like instructions for interacting with the operating system. We want to leverage the salient pieces with RAD tools to extend these applications.”
In the Blue Phoenix vision, organizations will be able to set up specific repositories for their own purposes, such as a development repository. “You don’t need every nuance of COBOL in that repository” to reassemble the business logic pieces into newer applications, Venema said. “There are only a handful [of COBOL assets] that are relevant to the regeneration process.”
The decision by Blue Phoenix to get on board with the OMG effort, rather than going in its own direction, is a good one. Many organizations today, which rely on mainframe applications but also see the need to reuse applications and move to service-oriented architectures, are experiencing pain because there is no standard way of achieving those goals.
A number of vendors, from application integration companies to legacy transformation companies, have attacked the problem from different angles. The OMG effort is the first industrywide attempt to define IT systems and build a model of the interactions. And Blue Phoenix, it appears, is ready to help OMG define the scope of its efforts. As Venema said, “OMG has always needed a vendor with a practical example” to lead the way.
David Rubinstein is executive editor of SD Times. |
|
| Data Watch |
Asia Pacific Development Survey: Do You Use UML for Your Application Design?
Click Here To See.
|
|
Advertisement
Click here for a complete listing of Industry Watch Columns
Click here to see a complete Column Archive. |
Back to Top
|