I stumbled upon a fascinating article in the June 2003 issue of Scientific American. I believe the vision set forth represents a much needed critical case study of today's situation, that being frequent breakdown of systems and poor reliability. Downtime is a much less significant problem to me than the actual restoration process which at times is a killer. "Self critical" systems should be the next step in the grand scheme of computing.

The project is labeled ROC for Recovery-Oriented Computing. The research team looks favorably to new concepts such as benchmarking the recovery time of a system, micro rebooting (that is restarting only subcomponents of a system rather than the whole system) and undo functions to trace the steps a system administrator would take.