Autonomic Computing Is Here, and It Works

 
 
By Ericka Chickowski  |  Posted 2007-11-26
 
 
 
Reaffirming its commitment to developing enterprise systems with self-healing and self-managing capabilities, IBM recently released a slate of new and updated Tivoli products with enhanced autonomic computing features.

Autonomic computing is a fundamental behind the creation of systems that self-regulate, much like the human body's autonomic nervous system, and it is the foundation for much of the work IBM has done on Tivoli since it published its Autonomic Computing Manifesto in 2001.

At that time, skeptics claimed Big Blue's autonomic computing vision was too ambitious. But, according to many IBM customers, the company has more than delivered on the autonomic hype.

Case in point: Network Solutions. The Internet services firm—best known for its domain registration business—openly credits its use of IBM's autonomic features with helping it navigate through two major acquisitions and save at least $1 million in IT operating costs.

"We believe in automation; we believe in systems telling us their health," says Jim Polkowske, vice president of operations for Network Solutions. "It's a philosophy and a discipline. If you think about the autonomics, we need a process and a definition that will guide us through change, make sure we can adapt to changing environments. We provide Internet services and we need to adapt to the Internet. We need to adapt to business change. And we need to adapt to new product changes."

Network Solutions uses Tivoli Management Framework, Tivoli Monitoring (ITM), Tivoli Workload Scheduler and Tivoli Enterprise Console. Polkowske says the tools enabled his IT department to create streamlined processes that take far fewer staff resources than conventional enterprise methods. Though the company's major line of business and its competitive advantage lies within its data center, the 1,000-person company only employs 100 IT members.

Polkowske credits Tivoli with helping to maximize IT employee productivity. Take software update deployments, for example. Before taking advantage of the automated tools, he had to have multiple staffers and subject matter experts at the data center to ensure no downtime was suffered during the process. Now all it takes is a button push to check system availability and health, check the code, sequence it, schedule it, and push it out to the systems.

"It was a very manual process and we had to have a whole series of individuals lined up ready to support it," Polkowske says. "Several weeks ago, I realized how successful we were with this not from the metrics we achieved but from the fact that we had a software release and it was only a question of figuring out who was on call that night."

In addition to streamlining processes like those, the automated computing features in the Tivoli products helps Network Solutions greatly diminish its recovery time during both major and minor network events.

Self-recovery plays out in fulfillment and billing activities, for example, with systems set to recycle or throw up troubleshooting alerts if orders per hour don't match the order history statistics. It also goes a long way toward helping Network Solutions deal with fluctuating Internet capacity demands.

"If we have one or two servers or a complete cluster go down, the system knows how to bring itself back up or to tell the other systems to fail over and become primary systems," Polkowske says. "On the Web services peer, we've built in the algorithm that we have to have x number of Web servers running at all times and we have our pool of servers that are busy doing other things. But should we lose a Web server or our capacity or threshold go up without any alarms, those Web servers in trouble will shut down and it will bring up the other ones."

The reporting capability also gives Polkowske and his team a robust system to analyze event trends and capacity levels to predict future needs. Not only do these predictive capabilities allow the company to maximize its uptime, but they greatly improve the accuracy of forecasts for hardware acquisitions.

"Where that saves us money is that it is now allowing us to figure out whether we need to allocate dollars Q1, Q2 or Q3 for funding or whether we can wait until Q4," he says. "It now allows us to make good, sound business decisions based on trend growth not only within our customer base but also when we look at Internet activity."

The tools were at the foundation of a successful integration of two newly acquired companies in the past two years, and in both cases the companies had relied heavily on outsourcing, according to Polkowske.

"We made the decisions to move them from outsourced to insourced models," he says. "From a support and IT perspective, the outsourced model for the operational support was just to add more bodies. Instead we said no, we can do it more efficiently with a good model."

The autonomic features, predictive capabilities and scalability of the tools made it possible to bring everything in-house with a leaner staff and fewer resources. They smoothed the way while migrating more than 450 servers managing 17 terabytes of data during the initial transition, he says. Polkowske estimates that Network Solutions has saved approximately $1 million in management costs through this move.

"Our success came when we were able, with very little effort, to take our current Tivoli toolset, extend it into the newly acquired systems and then move it into our data center with very little impact to our business and a substantial savings," Polkowske says, "because all of a sudden we've gone from a very manually managed process as far as IT support operations goes to a structured discipline with tools, central reporting and corrective actions."