Challenge: Convert bloated, NATURAL-based, legacy transportation-support system to Web-based front end without breaking the bank or working through the decade Solution: Hand it off to a couple of 20-something brainiacs who did it in several weeks for one-fifth the cost.

For this week?s CIO Nation Dossier, I spoke with BNSF Railway CIO Jeff Campbell about a project that blew my hair back. For years, BNSF had been pondering how to modernize its legacy transportation-support system and give it an online front-end that would make it more accessible, flexible, upgradeable, and cheaper to maintain. Here?s the amazing story of how they did it ? and how it can be a lesson for any CIO wrestling with the legacy monster?By Jeff?s own admission, BNSF had tried on several occasions over the years to address the issue of modernizing this system, which the company developed in the early ?90s to manage rail schedules, track inventory, and streamline logistics. The system processes 32 million transactions a day and resides on IBM System Z mainframes in tandem with DB2 data sharing. ?It was monolithic,? says Campbell. ?The problem is, it?s green-screen technology driving 3,200 command screens. The [newer employees] just want to point and click.?

Campbell said he wanted to convert it to a PC environment to broaden its usability. At various times over the past years, Campbell?s crew took a hack at modernizing the system, but the task was too daunting ? and expensive. His estimates ran into the $30 million range (assistant VP Bonnie Henn-Pritchard said she believes it would have been more).

About a year and a half ago, Henn-Pritchard, who runs the technical services team and reports to Campbell, gathered her folks for a strategies and innovation session ? a sort of brainstorming retreat. During that session, Henn-Pritchard put the challenge of converting the system to the team ? how to take, as she put it, ?a beautiful and elegant transportation-support system and make it easier to understand, faster, and more integrated.? Two of her younger employees, in their early twenties, took up the challenge and created a skunk-works project. Within six weeks, they?d developed a translator for converting NATURAL code into Java, and today the 3,200 commands in the old system have been converted into 60 to 70 Web-based screens. BNSF is now in the process of rolling out the first iterations of the new Web-based solution across the field. And, says, Campbell, the train-support system itself is ?still in Fort Knox, in the mainframe.?

Campbell says he was amazed at the speed with which these two employees worked (in conjunction with an employee of Infosys). After the six-week skunk-works project, it was another three months until the code was mature enough to bring to Campbell and the BNSF CEO. Campbell says that what would have taken about $30 million turned into a $6-$7 million project, adding that in his five years as the CIO, every project has had a required rate of return ? in this case, no one asked for a rate of return. ?It was just the right thing to do,? he says.

Campbell says the ramifications don?t end there. ?If we can do this, what about all that Cobol that still exists?? And, he hints, other enterprises can benefit from BNSF?s breakthrough.

Lesson learned? ?There is much great talent out there if just given the freedom, and the leadership, to [be creative],? he says. ?It?s a generational issue. My AVPs and directors are in their late 30s and 40s; I?m in my 50s. Then you?ve got these 20, 22, up to 30 year olds ? they look at technology through a different lens than we did. We?ve had people looking at this problem [for a while], and the only conclusion we?d come up with so far was to re-write the whole thing. These [newer employees] think in terms of SOA.?