HP 3000 migrations get compared to Y2K projects a lot, according to Speedware’s Chris Koppe. Not only for the complexity and crucial stakes of the multi-year efforts. When a migration project succeeds, users don’t even see a difference.
Koppe, who directs marketing for the e3000 Platinum Migration Partner, said his firm’s services team owns a 100 percent success rate in migrations so far, a period of work and research covering all of HP’s march to the end of its 3000 business. Staying perfect over more than six years boils down to three fundamentals.
“We leverage automated tools almost everywhere,” he said. “The next cornerstone is proven methodology and processes. The last one is resources — and if you don’t have enough deep knowledge of the legacy system, you won’t know what it’s supposed to do on the target platform.”
Speedware completed a migration recently of HP 3000 applications at Tufts Health Plan, an HMO running a mix of COBOL, PowerHouse and three dozen other technologies related to the 3000. At the end of a 30-month period, the HMO had 14 technologies running in concert on HP’s Unix, completed to move one batch and one online 3000 to HP-UX partitions mirroring each other.
Koppe said that Speedware is wrapping up four migration projects for 3000 customers this year. In a project that Tufts extended several times because of internal business reasons, the migration becomes “a non-event” to the company’s users, as invisible as any Y2K project.
The work at Tufts shared elements common to many such projects at a medium-sized 3000 site. A pair of N-Class servers hosted apps written in-house, with extensive utilization of NetBase replication and Omnidex optimization of TurboIMAGE. But the trick to success in these migrations is not mastery of the new technologies as much as melding the new mix. And the complete span of the necessary work doesn’t reveal itself on a first survey.
“We describe these as waterfall projects,” Koppe said, “where you’re not going to know everything that exists up front. You have to plan for a number of issues that will come up, and make sure your timeline has some flexibility in it.” Diving into the Tufts project revealed complex batch schedule dependencies, and “an application jumping between PowerHouse and COBOL at the user interface level.”
Migrations in the 3000 community usually mark a “code drop” as a fundamental milestone, “and the first code drop at Tufts was certainly a challenging one” because of the complexities. But most customers sign up for their share of the challenge to succeed, the portion they know better than any migration service provider: testing.
Speedware unit-tested its code for functionality, “and some customers want us to do all the testing for them. We have a very comprehensive testing workshop we do with the customers. It’s about a 50-50 split in terms of the work, because it’s not just IT people testing. Functional testing might be done by IT people, but user acceptance testing has to be done by the user community. The testing itself is very resource-consuming to an organization.”