The migration stakes fly high at WestJet
Using a migration general contractor's blueprint

Scoop up an array of migration savvy

A migration project can fail even if its budget is in the tens of millions. As our WestJet report relayed on Friday, $31 million wasn't enough to keep a co-developed effort from crashing at the regional Canadian air carrier. the company continues to rely on HP 3000 software that it's been using for more than seven years.

One of the ways to avoid this kind of crash is to make the best choices across the whole menu of migration entrees. WestJet, which is still hiring for 3000 help on an interim basis, was developing a replacement application alongside a single software supplier.

Not so effective, says Chris Koppe of Speedware. The marketing director and Encompass user group board member spoke at the recent e3000 Community Meet. He cautioned community members about searching for migration solutions in a single spot.

"I think its a poor approach to go to one vendor and have that one vendor build everything in-house, and do everything in-house," Koppe said. "There's a lot of great technology out there, and a lot of technology out there to automate different aspects of the migration."

Koppe was laying out strategy for a genuine migration — that is, the movement of code and systems from the HP 3000 to another platform, in the hope of retaining business logic and sustaining design continuity. The greatest challenge in these kinds of projects, he added, is the complexity of the typical 3000 system. Dozens of programs and utilities, usually, make up a reliable system.

  just about every customer migration at some point or another, and the average count of technologies on a 3000 is somewhere between 20 and 40. That's things like Suprtool, MPEX and Adager, data transfer tools such as ODBCLink or its full-bodied cousin UDALink, the bedrock layers like IMAGE and COBOL II. Each and every one has to be replicated on the new system, or if not, the migrated system must replace the functionality, somehow.

"When you're transitioning off the box the problem is not the COBOL migration, or the JCL integration," Koppe said. "The problem is the integration of all of these technologies." Speedware's got a list of 3000 technologies that it uses to discover what a migration customer has installed. It's a 200-technology list. There's more there on the 3000 than even a diligent manager might know about.

With a list that large, even the companies which use off the shelf solutions have unique integrations. "Two companies with the same technologies do not have the same migration challenges," Koppe said. "They've coded things differently, or used different intrinsics, implemented in different ways."

Koppe did offer a list of Key Tools for the do-it-yourselfer. Ordina Denkart's MPUX, the AMXW suite — there are a host of these automated migration aids — that a migration coordinator, the companies HP called Platinum Partners, assess and advise for a migrating customer. (For the record, there are only two remaining e3000 Platinum Partners from HP's list of four in 2002: Speedware and MB Foster.)

Even thought modernization offers more dazzling results in a migration, that kind of project is still rare among the migrating 3000 site. Koppe said Speedware's been surprised at the number of "Lift and shift" migrations, whose aim is to get the working applications and systems off the 3000 and onto any other platform. This is an aim that's "very low risk," he said.

"I think it's part of the 3000 community culture," Koppe added. "If you're still on a 3000, it's because you're not an early adopter." Chuckles followed that statement at the Meet. "You aim for business continuity, and technology that offers the least threat to the business."