Oracle launches servers for internet applications

Larry Ellison, Oracle's chairman and chief executive, has admitted that the database vendor's Business Online applications outsourcing scheme is likely to spin off into a separate company.

Speaking at a keynote speech at Internet World last week, Ellison revealed that Oracle will also manage the application hosting servers itself rather than outsource the business and will host third-party packages. While Ellison said that he expected 'thousands of ISVs' to take up the offer, he refused to disclose names.

But Ellison stressed that Oracle was not becoming an internet service provider, maintaining that customers would need to obtain internet access from another company because Oracle would only provide the server upon which the applications would be run.

The first Oracle packages to be offered under the scheme will be sales forecasting and order entry, and will launch later this month.

But competitors have pointed out that Oracle's approach is not original. Peter Wolcott, the recently appointed president of Pandesic, said: 'Imitation is the sincerest form of flattery.' Pandesic, which is a joint venture between SAP and Intel, has been hosting e-commerce applications based on SAP's R/3 application suite for several months now.

In a separate move, Ellison also hailed internet computing as the model that would kill client/server and rid the world of distributed servers.

The move was seen by commentators as a case of Ellison directing his venom away from the desktop to the sever market. Until recently, he predicted the imminent demise of the PC. He now forecasts the same fate for distributed servers.

While Windows-based client/server computing costs between $10,000 and $25,000 per user, per year, internet computing will cost no more than $1,000-$2,000 per user, per year, he claimed.

'If we can get the application off the desktop and back on the server where it belongs, wide area networks start to work again,' asserted Ellison.

The client-server model, which he referred to as an 'evolutionary dead end' is bound to disappear because of two fatal flaws: it is labour intensive and it requires local servers.

When applications run on the client, but data comes off the server, local area network speeds are affected and Wans do not work well, he said. This has led to a proliferation of smaller servers in remote locations, where they are difficult to manage and lead to the fragmentation of enterprise data.

'Client-server has devastated our ability to get information about our business. This model of Little Servers Everywhere is untenable.'