| About InfoWorld : Advertise : Subscribe : Contact Us : Awards : Events : Store |
 |
||||
|
||||
 |
||||
 |
|
|||
 |
||||
|
|
|
||
|
|||||||||||||||||||||||||||||||||||||||||
 COLUMN
Relativity Software CEO on turning Cobol code into gold Vivek Wadhwa, founder and CEO of Relativity Software, sees gold in the trillions of lines of legacy Cobol code that exist in many older enterprise systems. The gold is not just in the form of profits for his company, but in the ability for companies to reuse that code, and extract an extended return on as many as 20 years of costly investment.
How big of a task are enterprises facing in extending the life of legacy code? The rate of change of technology has started increasing dramatically, and the Web, [electronic] commerce, [electronic] business -- this whole concept of Internet computing -- has basically come like a tidal wave and hit businesses. There's a requirement that you be on the Web, that you have an e-commerce solution to your transactional system. And it's hitting the IT departments very hard. They're not used to having to adapt the systems so rapidly. The majority of the world's information systems today still run on IBM mainframes. If you consider that even 10 or 20 percent of these systems have to be moved over to modern platforms, that's a huge task. It's a massive marketplace. Why look to the past, to existing code, for modern applications? When client-server computing first came out, in the early 1990s, the first set of solutions were simply screen-scrapers to put content up on their systems, and they provided a tactical user interface. And then you had the higher-end development tools from the different vendors, which have basically seized the market. That was because putting a front end to a legacy system didn't really fit the problem. The ultimate solution was to rewrite the applications and have them run on in a distributed, client-server computing platform. Now, our solution, RescueWare 5.0, lets you take your entire legacy system and modernize it. That includes taking the user interface from an interface to an object, which runs in the Java environment. It includes taking the database from the proprietary, flat file structures over to the relational. And, most importantly, it includes taking the legacy code and turning it into true objects. In the legacy paradigm, you typically had monolithic programs that spanned hundreds of thousands of lines of code -- having a 10,000-line Cobol program wasn't atypical. They were very large objects. In the modern world, you are dealing with Java objects, which typically run in 100, maybe 200 lines of code. So what we do, basically, is transform the entire paradigm from the old to the new. How do you take older code and use it in a new architecture? What happens is that probably 40 or 50 percent of the code is obsolete because it deals with memory management and temporary storage cues. But the knowledge of the business is also embedded within that remaining code. Let's take a customer billing system. The business rules -- the method of doing the billing -- is embedded within the code. The truth is that the CIOs really don't know how the systems work. The people who wrote the code are long gone. So there is a lot of knowledge embedded within the legacy systems. We have this technology called knowledge mining in RescueWare that lets you go in and pull out those nuggets. You can transform the entire system and take it over to the new platform, or you can mine for nuggets of knowledge and transform those into the new application. We do both. How can you simultaneously take advantage of the old, in the form of transformed objects, while developing for new architectures? Let's say you're developing a brand-new system for an investment bank. You have systems that do bond transactions. You use our knowledge-mining tools to extract all of the business rules from the system and then turn them into Java or C++ objects. Then you can integrate those within your work. How does this relate to enterprise application integration (EAI)? Basically it allows you to mix in different types of application servers and different applications. If you need to integrate, say, PeopleSoft into your environment, you have to be able to understand what your systems do, and then pull out the key transactional logic from them and integrate them into your systems. Our tool can help you understand that and extract objects and then bury them into your new systems. And we're also writing up some new offerings specifically for that marketplace, but we're not ready to announce all of this yet. On a broader scale, what are some of the sights that you're setting for yourselves? We're working very hard to become a platform. We're working with the large system integrators, with the large software firms. We're going to be announcing deals with Ernst & Young in which we're going to become a part of their development center, and they'll start using it and modifying and adapting the tools to their own environment. So the strategy is to basically become a platform for legacy transformation. [We] see this as a multibillion-dollar marketplace, and we want to become the component supplier to this marketplace. So we're not typically going and marketing to end-users, to end-customers; we're typically marketing to the people who market to the end-users and end-customers. What is your view of how Java as a platform is moving forward? We're seeing it coming together very fast. I've seen more progress in the past 12 to 24 months than we saw many years before that, particularly in the past 12 months or so. I really see that Java is going to be a big winner here, and in the next one or two years, the key infrastructure problems, the key performance problems, [and] the key integration problems will all be solved. And it will become one of the major ways of developing systems. I'm very bullish on Java. InfoWorld Editor at Large <A HREF="mailto:dana_gardner@infoworld.com"> Dana Gardner </A> is based in New Hampshire. SPONSORED WHITE PAPERS 
SPONSORED LINKS
|
|||||||||||||||||||||||||||||||||||||||||
|
||||||||||
|
||||||||||
