We tend to think about autonomic computing in terms of enterprise, but this concept has traction in small and medium-sized businesses as well. Supernovas are explisions, hydrogen bombs about the size of the Earth. That’s pretty big stuff. We live in amazing times. If I think about Tim Malone’s talk this morning, those bands of people looked at the sky and didn’t even know what that stuff was. Neither did the kings, and some of our presidents.
That’s just one example of how we’ve become an information society. Half a billion people have access to the Internet. It’s moved into broadband, which is faster than the adoption of any other single technology. Technology absorption has been so rich and aggressive that corporate America spend almost $9 billion a year on spam, and that’s just one of the excesses. It’s had immense impact; look at the productivity numbers. Even Alan Greenspan agrees with me that at least half of that is attributable to information technology.
What’s bad about that? What’s the problem? As these systems proliferate, you introduce complexity. We used to think a bank’s ATM network was complicated. But think about how simple it was? Pretty simple stuff. Compare that to today. We have customers with 30,000 servers. Yet our processes and mechanisms have remained largely the same. Paul Horn, senior vice president of IBM Research, caught on to that point: “If we don’t get a handle on complexity, it will stop the expansion.”
We better figure that out. Then you look at what small and medium businesses need to do. 80% of small businesses have Web sites. They’ve got applications. They’ve got systems. Portable systems. They’re small scale compared to major banks and telcos, but it’s still a lot of stuff. They have no skill, so the complexity is even greater. How do we get a handle on it?
That brings us to the notion of autonomic computing. How many of you use the word “autonomic” in your day-to-day vocabulary? “Boy, when that bright light turned on, you really responded autonomically.” The vision is intelligent open systems that manage complexity, know themselves, continuously tune themselves, adapt to unpredictable conditions, prevent and recover from failures, and provide a safe environment.
We’ve broken the key attributes down. We think of autonomic systems as self-managing. You want systems that are more self-configurable and can adapt to dynamically changing environments. They need to be self-healing so they discover, diagnose, and act to prevent disruptions. You should be able to have operational efficiency by tuning resources and balancing workloads to maximize the use of IT resources. Finally, they need to be self-protecting to anticipate, detect, identify, and protect against attacks.
These might sound like futuristic goals, and it will take some time to realize them, but huge progress has been made. It’s not just hardware progress. Database administration now includes learning optimization, recommendations of automated summary tables, object management, and a broad range of things. Notebook computers can now have an accelerometer built into them so they can tell if they’ve been dropped and put the drive mechanism into lock mode.
The cornerstones are in place. The concern, though, is that doing individual improvements won’t add up to the promise of a system that has these organic, cohesive characteristics. We need a holistic solution.
This is all based on the concept of sense and respond and control theory, which is big in manufacturing. Using sensors, we can monitor, analyze, plan, and execute changes in a biofeedback loop. We need to build intelligence into our systems, but we also need to build interfaces that make them part of the whole.
One of the ways we can do this is through virtualization. Virtualizing servers, storage, and networks simplifies the infrastructure and reduces management complexity because you pool all your computing resources. How might the orchestration of these resources work? Traditional infrastructures brought low overall resource utilization and high labor costs. Orchstration increases resource utilization, reduces labor and cost, deploys resources dynamically, and aligns business and IT priorities. Most organizations might go with a hybrid model.
This has been put to work for the Australian Open. We started this at the U.S. Open last year, and it’s gotten more sophisticated. The Australian Open is a two-week tennis tournament, and on their Web site, they track every stroke, every match. A single infrastructure simultaneously manages three unrelated networks. Server resources shift automatically to manage fluctuating demand.
Big steps are being taken, and the best and brightest are exploring autonomic computing. People are looking at predictive optimization, policy negotiation and conflict resolution, context awareness, the human-computer interface, cultural change and trust, and other topics. But the journey has only started. Industry-wide collaboration is essential. Innovation is required. And we need more information and tools available. No one company can do this.