There have been some strange and wonderful software products targeted at the enterprise software customer over the years, but few have been quite as intriguing as Huron, an ambitious, multifunctional and much misunderstood development tool launched in the early 1990s.
In spite of a customer base of more than 100 organisations globally, few modern programmers, or indeed IT directors, know anything about Huron or the remarkable ambitions of its Canadian developers.
But now, a decade on, a newly formed company, Objectstar, with an eponymously rebranded product, is working to bring the dream back to life. Its problem will be to try to overcome some of the same issues that dogged the product’s progress so many years ago.
When it first launched Huron in 1992, its then owner, Amdahl, a mainframe manufacturer with little history of software development, seemed unsure of how to present the product.
Huron was not, it said, an application development system per se, although it contained its own scripting language and tools for development. It was not a fourth, or even a fifth-generation language, though it might compete against these; nor was it a database, though it contained its own datastore; nor an integration platform, though it could tie applications together; nor was it some kind of transaction processing system, though it contained one of these. It was, the company explained, an ‘environment’ for capturing and managing business logic.
And it wasn’t just any environment. The whole thing, which had been conceived by a Canadian developer in mid-1985, had taken six years to bring to market – even building the prototype, with millions of lines of code, had taken three. It was, the company argued, exactly the kind of flexible system that project managers, especially at large sites, badly needed to solve their serious development and deployment problems.
But Huron never lived up to its promise. For a year or two, orders were signed and customers gave glowing testimonials. Then everything changed: IBM gradually clawed back control of the mainframe market, forcing Amdahl to sell its remaining shares to Fujitsu, and then to pull out of hardware altogether. Huron, for a while, was used as the basis of Antares, a services alliance with EDS, but this fizzled out, leaving Fujitsu unsure of what it had. Investment stalled.
The problems weren’t all of Fujitsu’s making. When Huron was launched, some 80% of IT software budgets went on development, prompting a group of companies, such Seeq, HPS, Seer, Forte and Dynasty, to enter the market with products only a little less wonderful than Huron. But then came new client/server tools and the Internet, the Y2K milliennium bug and the giant business packages such as SAP. Large-scale, high-end development spending dwindled.
But history has a strange way of repeating itself, says Dave Hoyle, now chief technology officer of ObjectStar, which split from Fujitsu in April 2003 with the help of an undisclosed investment from US venture capital company InvestCorp. He believes IT directors are ready to spend to solve their business process problems.
But ObjectStar executives find themselves with some familiar strategic and marketing conundrums: How should they position the product in what is, once again, a highly competitive market? And how should they develop it?
Even today, it is difficult to categorise the product. Jim Hoffmann, CEO of the company, knows it is important to get this right. Parts of ObjectStar compete with IBM’s WebSphere, with BEA’s WebLogic, and with development environment’s such as IBM’s Rational Rose.
But at the same time, it could just as well be complementary. CTO Dave Hoyle also describes how the vision for Huron in its early days resembles a business process management (BPM) system – which even today only a handful of vendors claim to have developed. “The architect of Huron, Helge Knudsen, came up with the idea of an environment that would enable change. The problem was not so much deploying applications as changing them.”
His solution was to extract the business logic and store the rules in a meta-store. Changes are made through a scripting language that does not compile down the low-level changes. In this way, it resembles the Java virtual machine.
While ObjectStar is in many ways a modern environment, its age is both an asset and a concern. Its longevity means it is proven and scalable, with referenceable customers; but it also means it was not written in a modern, open way. A big effort to include open standards, such as java-based adaptors, has overcome some of this, but some buyers may still baulk at the proprietary scripting language.
Hoffman says that initially, ObjectStar will be targeted at the integration market with an environment for linking applications, especially legacy ones, at a process level. But eventually, it will be positioned as a high-level process development environment.
ObjectStar won’t reveal its sales, but says they are above $10 million – with most now coming through channels and outside the US. And it is profitable. With customers, and profits too, that is a good start. But it is going to get very crowded in this market, with IBM, BEA and Microsoft determined to win the biggest share.