IBM on Wednesday said it was teaming up with two U.S. universities on initiatives that the company hopes will improve the self-healing and self-managing features of physical and virtual servers in a broad-scale computing environment.

The company will team with Georgia Institute of Technology and Ohio State University to train students and researchers on software development that enables effective management of system resources and prevents jeopardizing of large-scale computing environments when a critical service is taken down.

IBM will also focus on the development of self-managing software that addresses the balance of system availability and performance in such large-scale environments, said Matt Ellis, IBM's vice president of Autonomic Computing.

For example, if a server or virtual partition needs to be updated in a large computing environment, a part of the research could lead to the development of software that addresses infrastructure changes on the partition shutdown, Ellis said. Software from the initiatives could recognize and move workloads to another physical server or virtualized area to keep the system running without downtime.

First results from the partnerships may appear in 12 months, an IBM spokesperson said. IBM will work with the universities to see how this research can be realized into products in the first 24 months, the spokesperson said.

In a broad-scale computing environment, which IBM refers to as "cloud computing," data and services reside in massively scalable data centers that can be accessed from connected devices over the Internet. The demand for such environments has been driven by the growth in recent years of technologies like social networking, streaming media and mobile Internet devices, Ellis said.

Companies including Amazon.com and Salesforce.com have invested in cloud computing.

Although IBM sees cloud computing as a cost-efficient way to provide applications and information management services, the research is important to keep up with the rapid changes in broader, more virtualized environments, Ellis said. As computing environments become more virtualized, there is an increased speed in which system changes are made, he said. A physical data center has limits on how fast new systems and storage can be added.

The IDG News Service is a Network World affiliate.