How to Lay Dead Technologies to Rest

Inside every company live systems that, although built on old technology, are core to the business and can’t be migrated easily to more modern software or hardware. That’s the nature of IT. Build and install an application today, knowing that in five or eight tomorrows, the technology will be outdated.

More on Legacy Systems

Technologies We’re Glad are Dead

Crash of Legacy Systems at Comair

Project managers try to choose technology they think won’t get obsolete quickly or one that future products can link to. Crystal ball stuff like that is tough and no one predicts accurately all the time. Even today’s modern SAP or Oracle enterprise resource planning suite will be a legacy system someday.

Disney is trying to come at the problem another way: making business and IT people involved in project planning aware that, eventually, the technology they’re using will have to be replaced. So that at the same time they’re planning what to build, the project team can also plan how and when to take it apart.

“I tell business people that putting in a system is like pouring concrete. You’re going to need a lot of jack hammers to get it out someday, and this is the opportunity to plan for that,” says Steven P. Davis, chief architect and vice president of IT at Walt Disney Studios, a $9 billion movie company that’s part of the $34 billion Disney entertainment empire.

Davis wants to classify every IT product at Disney as emerging, under evaluation, core or declining. “Domain leads” inside Disney—subject matter experts in, for instance, servers or application software—are assigned to write research papers about their technologies, to identify what’s declining. These papers are then released in the fall, a couple of months before Disney’s annual business planning cycle begins in December, Davis says. Davis spoke at the Society for Information Management‘s annual conference in Memphis in October.

That’s enough time to get it in front of the planners before project decisions are made, he says. Some technologies, such as server software, are assessed annually. But others are moving so fast that quarterly reviews are the norm, for example, BlackBerrys, other PDAs and smartphones.

Once hardware or software is labeled declining, something must be done about it.

One rule Davis has is to keep within one version of the latest iteration of a piece of software. For example, if the most recent full version of Microsoft’s database is SQL Server 2005, Disney will run no more than one version behind—SQL Server 2000. When SQL Server 2008 arrives, expected this spring, Disney, presumably, will step up the trudge migrating its databases from 2000 to 2005.

But just because a technology is identified as declining and earmarked for replacement doesn’t mean it happens instantly. One set of declining servers—Davis won’t name the vendor or model—has been undergoing replacement since 2003. He started with 27 and is down to four, he says. Finding time—convenient time—to migrate the applications running on the old hardware servers to new ones is hard when the systems are critical, he says.

“It took four years to push ’em out the door,” he says. But, he adds, he kept the old machines to his “one version behind” rule all the while.

Disney’s criteria for assessing whether a technology is dying:

• Your company is migrating away from it or no new implementations of it are planned
• No longer supported by the vendor
• Costs to maintain the technology are escalating
• Unsustainable, meaning it’s hard to find people who know the technology or, when they materialize, says Davis, “they hold us over a barrel for a lot of money.”
• Out of favor in the market, perhaps replaced by a newer version or model
• The vendor is risky—financially unstable, for example, or likely to be acquired. “We have to be very careful about our brand,” Davis explains. “We cannot have a supplier that’s going to put us on the front page of newspapers” should it go bankrupt, for example.