In the January 2010 IEEE Computer magazine there is an article Fulfilling the Vision of Autonomic Computing (unfortunately only the abstract is available for free). This articles looks at the progress that's been made since 2001 when IBM published an article on the need to automate computer management.
This new article discusses how in 2001 there were many unfilled jobs in the IT space and that at the rate things were targeted to grow, we'd never keep up and how clearly in 2010 that is not the case. There are many ways that we manage computers. We use virtualization techniques on the small scale today, things that were only done on the largest scales in 2001. It isn't the same autonomic approach that IBM proposed but it was "good enough".
On the other hand, we do have much more power and data that can be applied, and if it can be applied to the computing environment it can be applied to the enterprise as a whole. When we overcome the relatively arbitrary distinction between the business and the IT departments and address the problems from an "and not or" perspective, it is inevitable.
The IEEE article talks about a couple of examples where autonomic techniques are gaining traction:
"The main cost for the operator of a data center is power, thus provisioning of systems to match the workloads and service-level obligations becomes a critical business success factor. Because workload demands change minute by minute, no human operator can provision services with sufficient efficiency"
And
"Applications like environmental sensing cause the network to meet the real world in ways that preclude direct human management. The viability of environmental sensing - essential for efficient science and policymaking - therefore depends on sensor systems' ability to self-manage in the face of a changing environment"
If you change the first example to "plant floor" instead of "data center" and "enterprise" instead of "network" in the second example, it is equally valid. As we increase our abilities in IT to use these techniques, the real high value return will be to apply them for the enterprise as a whole.
Granted we'll not eat the elephant whole, but I'm sure there will be more autonomous progress for the enterprise in this decade than there was for computing in the last.
