Mapping a New Harbor for 3000 Data

Michael Marxmeier is creating a new harbor for companies who must sail out of their familiar HP 3000 database waters. The most recent efforts of his company have delivered a version of Eloquence — the database built for non-3000 systems that works like the 3000’s IMAGE — with value and speed similar to the database at the 3000’s heart.

Marxmeier, 43, has spent more than one-third of his life creating and nurturing Eloquence. His company based in Wuppertal, Germany has seven full-time employees, and after retaking full ownership of the database from HP this spring, he says it’s ready to grow even larger. The size of Marxmeier Software AG becomes an issue when customers match it against the databases from competing companies: Oracle, Microsoft and IBM, offering Oracle, SQL Server and DB2. HP 3000 customers have long seen small software companies deliver far better product and support than larger firms, however, and Eloquence has an advantage over the big names: it sells for little more than $7,000 per server, far less than the better known products. What’s more, developers report that Eloquence includes an awareness about IMAGE intrinsics that no other database even tries to match.

Like so many of the HP 3000’s developer gurus, Marxmeier reports he got hooked on computers during high school. Also like some, he started in a college program and then found the business world more interesting than completing his computer science and math studies. His computer career began working for an HP reseller, introducing him to HP systems. Those were HP’s Unix systems, however, the computers for which he built the first version of Eloquence in 1987. He’d already built up a store of IT business experience, managing three different branches of a medium-sized business’s computer operations at age 22.

In person Marxmeier still carries a youthful enthusiasm. At last year’s HP World I watched him verify his discoveries about the internals of IMAGE with HP’s Tien You Chen. His informal conversation that I overheard sounded like a veteran speaking of HP 3000 database experience, as Marxmeier explained to HP’s top IMAGE expert his understanding of the database’s internal workings.

Eloquence was born of an HP transition similar to the one the 3000 community has begun. The database was at the heart of another business system HP stepped away from, the HP 250. That small business machine had a customer community much like the 3000’s and Marxmeier wrote a work-alike system to help his company move its applications onto HP’s Unix systems. Forced to leave a well-designed environment, the experience seems to have given the company a humility in its 3000 transition mission that stands in sharp contrast to the swagger of Oracle and the other alternatives. It seems that the Eloquence alternative offers more for the typical HP 3000 shop that has learned to be efficient with its resources.

Eloquence was sold to HP at first, but this year the software and the future of the product returned to his ownership. This step away from HP impressed as an independent move, one of the few things that seemed to stand in the way of a broader future for Eloquence as an IMAGE replacement. Marxmeier seems to direct the development with an eye toward possibility, already moving his software onto Itanium systems earlier this year. He moved Eloquence to Linux in 1997, and believes it was HP’s first software ready for that environment.

In that same year the company took over sales and distribution of the product, but this year marks the first that this small company is truly out on its own. We wanted to know where Eloquence can take 3000 sites with IMAGE applications that need to be preserved for their business rules — and where Marxmeier wants to take the database that looks destined to harbor HP 3000 data.

What part of your business did HP 3000 customers represent 10 years ago? Is it fair to say nearly all initial customers were not MPE users?

Eloquence was originally created to move HP 250/HP 260 applications to HP-UX. The HP 250 was a low end business system, typically used for small business or branch offices and was discontinued in 1990. The HP 250 was a success in the US and did extremely well in Europe. You can find a few notes on the Eloquence Web site.

The huge majority of the HP 250 VARs and installed base moved to Eloquence in the early 1990s. Eloquence managed to make the transition painless by providing backwards binary compatibility. A number of current applications based on Eloquence have their roots on the 250.

There used to be little overlap with the Eloquence customer base and the HP 3000, because the systems were aiming towards different markets.

What made you continue to work on TurboIMAGE compatibility for Eloquence before November, 2001?

We initially got started with TurboIMAGE compatibility in a migration project with a well known HP 3000 ISV. They investigated options to migrate their application to HP-UX. Since we already had a similar IMAGE implementation available on HP-UX we became part of that project. It worked well and we decided to continue with this project.

When we started in 1999 it seemed to be an easy task. Our IMAGE implementation was working for years and the difference did not seem that big. With some hindsight it’s safe to say that its always the remaining 10 percent of the work that takes 90 percent of the time.

How much time have you devoted to learning IMAGE at a detailed level?

It depends how you look at it. Anything between a few weeks up to 25 years. I started with IMAGE on the HP 250 in 1979. We implemented the first HP 250 IMAGE workalike on HP-UX in 1988.

While the HP 250’s IMAGE differs from TurboIMAGE, both follow the same approach. The TurboIMAGE implementation has its own history and details that applications rely on. In the development process I learned a lot about IMAGE beyond the documentation.