The Legacy Systems. The architecture plan will serve as a map for Cooper and his team. But this is no greenfield exercise. To create a department capable of securing the US, they'll have to integrate hundreds of legacy systems. The first step is an assessment of the 22 agencies' infrastructures, which itself is a hugely complex task. There are 22 personnel and seven payroll systems among the agencies, for example. You can bet that a chunk of those will disappear in DHS.

Cooper and Flyzik are also building an inventory of the current enterprise licences, with an eye on streamlining that number. The number of applications that deal with homeland security is around 500, according to Flyzik. Some may be dropped if they're redundant; the rest will need to work together. And new ones will likely be added too.

Last July, the feds showed they were serious about integration when the OMB created a supergroup of federal CIOs to pave the way for DHS's effort. The IT Investment Review Group - made up of Cooper, Flyzik and the CIOs of the Coast Guard, Customs, Federal Emergency Management Agency, Immigration and Naturalisation Service, Justice Department, State Department, and the Transportation Security Administration - was charged with reviewing government IT investments to identify redundant spending and decide which assets could form the basis for integration efforts.

John Koskinen, city administrator for the District of Columbia and President Clinton's Y2K czar, thinks that the Y2K exercise has put the government in a much better position to pull off integration. "Everyone has a much better inventory of the systems they're running and why they're running them," he says.

But the reality of the integration challenge is this: there's a ton of legacy stuff in those agencies, and connecting it could be messier than a xxxJack Welch divorce proceeding. "A lot [of systems] will be old, built out of technology that doesn't exist commercially any more," says Raymond Wells, the former CTO and CFO of Alabama who is now director for strategic transformations in the application and integration middleware division for IBM. He cites the FBI, one of the agencies that will be sharing information with DHS, which has to send photos by overnight mail because its e-mail system can't send them digitally.

Cooper and Flyzik plan to overcome the legacy hurdle with middleware, EAI tools and the Holy Grail of integration, Web services. Yet there's a lot of disagreement in the tech world over when Web services will be ready for prime time. Yes, the major software vendors are betting the integration farm on Web services, but the latest and greatest of the overhyped technologies still needs to grow up a little. Much work remains to be done on security and the reliability of transactions, in particular.

BARRIERS AS BIG AS A CONTINENT

A more daunting challenge than actually connecting the agencies in DHS may well be how workers deal with the humongous amounts of data flowing into their systems. In addition to the information streaming into people's desktops from within DHS, there's another huge spigot - the FBI, CIA and other intelligence agencies - that will be bursting the bandwidth with gig upon gig of data. Then there are the state governments, localities and private companies, which will be sharing information in areas such as law enforcement and intelligence with DHS and vice versa. "It may be the largest volume of information ever," says Rock Regan, CIO of Connecticut and president of the National Association of State CIOs. Add the fact that much of the data will be dirty, redundant and useless, and you've got an analysis problem the likes of which has never been seen.

The Knowledge Glut. "The difference between data and information is the Achilles' heel" of DHS, says David Colton, vice president for strategic initiatives at the Arlington-based Information Technology Association of America (ITAA). "The really important thing will be to sift through the mounds of crap to figure out what's important . . . If you get all the operating systems working together but don't turn data into information, you've still failed."

"It's a fascinating intellectual challenge," says Koskinen, who notes that intelligence agencies have long wrestled with this issue - for example, filtering data intercepts. "The infrastructure issues are challenging, but you can deal with that. Much harder is, What's my system for sorting and filtering and analysing all that data? If [DHS] has reams of data and no one can figure out what it all means . . . you're talking about people's lives all over the world," he says.

The fact that the intelligence agencies remain outside DHS makes information sharing that much more hazy. The FBI, CIA, National Security Agency and others will still be the main gatherers of intelligence information. The plan is that they will pass on intelligence to DHS. That's not a good model in Colton's eyes. "If [DHS is] receiving finished analytical reports, they're a passive customer. That doesn't work; it doesn't solve DHS's needs. When the intelligence community is reformed, DHS needs to become a much more dynamic part of it. Absent that change, I'm not sure the department will fully succeed," he says.

Cooper and Flyzik believe the concept of "capture once, reuse many" (capturing data only one time to eliminate redundancy) will help them confront the spectre of too much information. As part of their architecture work, they're identifying and labelling "databases of record", which will then be designated as official sources of data. Other databases would be cast aside; new ones could be created. "We recognise it's probably going to take a few years to get where we want to be fully," says Cooper.

In many ways, this is the mother of all knowledge management projects. And there isn't exactly a font of KM successes in the private sector to use as a model. It's hard to find people with broad experience in labelling, organising and retrieving information, notes Woody Hall, CIO of the Customs Service. He also cautions that DHS will need to be careful in its mission to provide information to people anywhere at any time, so that the end result isn't a push kind of architecture, in which people would be inundated with data and information. "The challenge is to sift and screen it," he says.

Flyzik is adamant that DHS workers won't suffer from information overload. "If you get the architecture right, the net result is you will get the right information to the right people all of the time," he says.