Systems that look after themselves

Autonomic computing is an attempt to do for software, networks, servers and workstations what the human autonomic nervous system does for our bodies, according to Kathy Mandelstein, program director at IBM.

IBM has high hopes for this technology -- which combines tools developed by its research center and Tivoli unit -- that in the ultimate scenario would allow computers to diagnose and fix problems with little or no human intervention. The job of the developer may change somewhat if baked-in autonomic fault tolerance becomes a widely accepted systems goal.

"It's very similar to the autonomic system we have in our bodies as far as making systems that are self-configuring, self-healing, self-protecting and self-optimizing," said Mandelstein.

Self-healing involves tools that run periodically, do diagnostics, send alerts to system administrators when trouble is detected and then prompt them with suggestions for correcting the problem.

"In some cases, scripts would allow the network to go ahead and make corrections itself," Mandelstein said.

If all this sounds a bit futuristic, it just may be. A lot of software has to be developed to help that hardware run silent and deep.

Early adopters of autonomic computing include IBM, which uses tools to monitor and diagnose its own systems, as well as some of its larger customers, Mandelstein said.

To help developers learn more about the technology, IBM has created an Autonomic Computing Zone on the developerWorks Web site. The zone, launched this month, includes information about autonomic computing, from introductions to tutorials and user scenarios.

It may be the case that a lot of best practices forged in other software fields need to be adapted within the emerging autonomic model.

If you view an introductory matter piece on the Common Base Event model described by IBM Java/WebSphere Trainer David Bridgewater on the IBM Autonomic Computing Zone, you find that machines are similar to humans -- for analysis tools to understand an application's messages, those messages have to be in an expected form using expected terminology, writes Bridgewater. IBM and others are applying XML meta data means to this end, and are working to standardize approaches that have been a mix of standards and vendor-grown approaches to date.

While developers are already aware that their error messages must be utilitarian, they may find that their creation of error messages will have to fit more precisely in a special framework if autonomic computing rolls out of the lab and into the workplace.

For more, please go to the links: Autonomic Computing Zone on the developerWorks Web site found at

"Standardize messages with the Common Base Event model," a developerWorks article found at

About the Authors

Jack Vaughan is former Editor-at-Large at Application Development Trends magazine.

Rich Seeley is Web Editor for Campus Technology.


Upcoming Events


Sign up for our newsletter.

Terms and Privacy Policy consent

I agree to this site's Privacy Policy.