Bosses debate Net future

In San Francisco last week database company bosses discussed the impact of the Internet on the future of database development. Stuart Lauchlan was there.

The Internet is changing the face of the database market, but the chief executives of three leading suppliers have different views on how to exploit the shift in application development that will ensue.

The second session at the Planet Wall Street Conference in San Francisco last week tackled the relationship between databases and the Web. Three database chief executives - Sybase's Mitchell Kertzman, Informix's Phil White and Object Design's Robert Goldman - took part.

At first it appeared the session would be characterised by a consensus uncommon in the typically combative database market, as all three bosses agreed that the Internet changed the nature of their businesses and the way their customers used their products.

Goldman, a 25-year veteran of the database industry, was quick to point out that this was not the first time the market had gone through a period of major upheaval as a result of a new technology.

In the early 1980s, Goldman worked for Cullinet, a major player in the network database market which was swept away by its inability to react quickly enough to the advent of relational technology.

'The Internet is driving the next major shift in the whole computing architecture,' he said. 'It's been called a lot of things, from distributed applications to distributed computing or distributed object computing, but it's a fundamental shift to a distributed computing architecture from a client-server architecture.

'It is enabling new technologies to emerge. We see Netscape emerging as a software infrastructure player. It's also allowing new database technologies to be needed.'

Sybase's Kertzman took a slightly different tack. While not disputing that an Internet-driven shift was underway, 'when the technical tectonic plates of the industry move', he argued, this would take a different form and would potentially have different results.

'In the past, we only knew that a paradigm shift had happened in the way we know that an earthquake has happened,' he claimed. 'We notice its after-effects. I think we're in the process of the first potential shift which is being promoted ahead of the shift.

'People have recognised the potential of paradigm shifts for transitions of power and the creation of wealth. There's a great deal of desire on the part of a lot of people to have (the Internet) be such a shift.'

Previous technical revolutions have claimed the life of several of the established players in the market and spawned a host of new companies in the process. It wouldn't happen this time, claimed Kertzman.

'One of the signs of a real shift is that the big companies cannot adjust,' he said. 'But if you look at Microsoft, Oracle, Sybase, Informix and other companies, they're adapting pretty well and in many places taking a leadership position.'

This raised the contentious issue of the best data model to exploit the Net and Web-based application development - relational, object-oriented or object/relational?

Kertzman defended the relational model. 'A lot of the very high volume, very dynamic Web sites work very well on the relational model.

'Stuff like CNet and HotWired are running on Sybase relational databases today and running pretty well,' he insisted, adding the proviso: 'That's not to say that we don't need object extensions - we're working on those too.'

In the light of his company's high-profile launch of Universal Server in December, Informix's White was predictably robust in his pitch for the object/relational hybrid model. 'Customers have invested a lot of money in client-server computing and they don't want to throw the baby out with the bath water,' he said.

'I'm a firm believer that object databases aren't the end-all. The extensions to that, because of performance and recovery and everything else that's in the market, are exactly what people want.

'I have yet to see a successful object database in application inherently deployed and running in large scale. What we've said is to migrate the relational model to the object/relational model without writing a single line of code. As you want to add new richness to it, plug it in with DataBlades and get on with your life,' he explained.

Goldman also argued that application development itself has to change.

'People need to implement applications very quickly - if you don't have a site up there they'll go elsewhere,' he commented.

'The first customer for Time Warner's Pathfinder sites was CompuServe with three million accounts', he said, by way of example. 'So when they went live, they instantly had three million users. To design and build that system in a way that would support (those numbers) is an absolutely critical issue. It's a very different model of building applications than we've seen historically.'