Researchers See Steady Focus On Utility, On-Demand Computing

More vendors are investing to leverage future technologies that use on-demand infrastructures and utility computing, several top industry researchers said at a conference in New York in mid-November.

Research executives from IBM, Hewlett-Packard and Intel all described their efforts as customer-intensive, utility-based efforts that will roll out gradually in new products over time.

"We do a lot of customer work through IBM Research," said Alfred Spector, vice president of services and software at IBM Research. "That drives our thinking. When customers get interested, product groups get interested very quickly."

Spector said IBM continues apace with its efforts to roll out on-demand computing technology and pursues new functions in natural language and autonomic computing, among other features.

Rich Friedrich, director of the Internet Systems and Storage Laboratory at Hewlett-Packard Lab's Internet and Computing Platforms Research Center, said his company is also performing R&D on utility computing models."The goal is to make IT infrastructure more flexible, as well as more cost-effective," Friedrich said. "We're trying to increase the flexibility of the IT infrastructure to allow it to respond to changes."

In one instance, HP Labs recently worked with a British media company, 422 South, to provide a utility, on-demand computing environment for rendering of animated videos. What would have taken nine months to render from story board to DVD took 17 days with a utility computing infrastructure that made computer capacity available to 422 South on an as-needed basis.

"[Now], they'd like to buy up service and when they have excess capacity, sell it back to the grid," Friedrich said.

Researcher Mahadev Satyanarayanan, director of Intel Research, Pittsburgh, and a professor of computer science at Carnegie Mellon University, told the conference his organization is developing an Internet Suspend/Resume technology that provides a desktop portability not using physical hardware, but by preserving data on parcels maintained on servers and delivered over the Internet.

"All of your computing is done at the machine," Satyanarayanan said. "You have that state saved on back-end servers. By using a complete saving state on the back end, it reduces the total cost of ownership tremendously."In that scenario, he said, the primary hardware is maintained centrally within an enterprise. Individual clients have their data stored as a packet that can be retrieved from any location. All applications, middleware and hardware are maintained and upgraded centrally by a single IT department.

Intel research is in the process of rolling out a pilot program with 300 users to analyze their patterns, habits and how they use such a portable, utility, on-demand client computing model, he said.

With the heavy emphasis on integration, Friedrich said HP is mindful of the role its partners will have in deploying these computing environments.

"We need to make sure there is a strong ecosystem for partners in that environment," he said.