Hewlett-Packard Corp., Oracle Corp. and Intel Corp. are plying a middle way for service-oriented architecture (SOA) to replace mainframe COBOL applications.
IBM, which is the vendor for most of the mainframe hardware and software that falls into this category, still insists that the mainframe can be a viable component in SOA. Recently Dave Locke, director for IBM Rational worldwide marketing strategy, said: "There's tons of good working programming on mainframes. Almost all of it is in COBOL with some in PL/1 and other languages. There are some 200 billion lines of COBOL. It's been said that if we replaced all that COBOL it would cost $20 trillion."
But rivals HP, Oracle and Intel with their Application Modernization Imitative (AMI) are not planning to replace all 200 billion lines of COBOL, just the business critical parts that can become part of an SOA implementation.
"SOA and application modernization are joined at the hip," said Paul Evans, worldwide director for application modernization services at HP, who is spearheading the AMI initiative, which includes "offloading from a mainframe environment to an HP environment" with Intel processors. AMI also includes Oracle Fusion middleware.
"A win for us in the Application Modernization Initative has to have components from Intel, HP and Oracle," Evans said. "It can't be just two out of the three."
This past month, AMI announced three such implementations, which Evans said gives credibility to the initiative first announced at the 2006 Oracle Open World.
AMI is now being adopted by three European organizations the Swedish Tax Agency, BBVA, a Madrid-based multinational financial services group and Carrefour Italy, part of one of the world's leading distribution companies and the world's second-largest retailer.
"The three wins are in Europe, but we also have business prospects in other parts of the world," Evans said.
These AMI adoptions are part of a larger movement of corporations that want to do SOA, but want to do it in "tactical steps," he explained.
"When we go in a talk to customers about modernizing, we also say to them whilst in the process of modernizing, we are assuming that we are going to head in the direction of SOA," Evans said. "Basically the whole modernization thrust that's going on in the market involves people who want to move to the future state called SOA."
While mainframe advocates say they can play in SOA, Evans argues that is dangerous thinking. While COBOL programmers are still available, he said computer science graduates lack both the interest and the skills to work in the 1970's language.
"Just leaving it in COBOL is a slippery slope," Evans said. "You are dealing with code that is not flexible, not agile, not aligned with the business process. You're probably dealing with an architecture that's 10-, 20-, 30-years old. You're spending a lot more money on processing that code using a mainframe environment," he said.
However, he is not advocating throwing the business critical baby out with the COBOL bathwater.
"First of all," Evans said, "you've got to analyze what's on the mainframe because there are certain apps that are on the mainframe that would give you very little benefit if you took them off. So you've got to decompose what is on the mainframe because you're not going to take every single app off. These mainframes are running hundreds and hundreds of apps. It's too risky and it's not necessary to take them all off in one go."
The AMI initiative starts with an analysis of what code is on the mainframe before consultants advise re-writing it in Java so it can better fit into an SOA implementation.
"So what we try to do is identify the ones that would give you significant benefit, but are also set up well to be taken off the mainframe," Evans said. "We identify those and then start the process of identifying a future environment which they are going to."
For the risk adverse IT managers, the AMI initiative offers re-hosting, which is basically moving the legacy application from an aging mainframe to Intel-chip-based HP servers.
"It comes down to once you've analyzed the app, what makes sense for the customer both economically and functionally," Evans said. "Some customers are quite rightly very risk adverse. Their view is they want to take small steps first, so they may want to re-host, which does not give them a big step towards SOA."
But more innovative organizations not only want to move off legacy hardware and re-write some of the legacy code, they are focused on moving to SOA, he said.
"Organizations looking for application re-engineering, which is effectively decomposing and re-writing applications," Evans said, "these are the guys with asperations to SOA."
The three vendors supporting AMI do not push legacy rewrites and conversion of every application to Java," he said.
"I'm not saying running things on the mainframe are totally wrong," he said. The consultants working on AMI offer to analyze the legacy applications to give the IT managers a view of where they stand.
"When we looked at the COBOL code and decomposed it there are situations where clearly there are parts of the application where we say we don't think it's worth moving that at all," Evans said.
If an organization is focused on flexibility then Evans and his consultants recommend re-engineering the COBOL applications into services that will work in an SOA environment, he explained.
That is the direction Evans sees organizations headed whether they are taking tactical steps or re-engineering an entire IT system.
"Everybody is looking at this," Evan said, "it's just some people are moving quicker than others."