Fault-recovery and intelligent distributed system
Zbigniew W. Raś · 2002
A site of a distributed system can acquire new knowledge through learning from its closest neighbors. This new knowledge, stored at that site, is in a compressed form and is presented in the form of rules. In this way each site of the distributed system has some picture about the knowledge kept by its neighbors. The proposed scheme is characterized by fault-recovery and fault-tolerance capabilities. If one of the sites of the distributed system is somehow destroyed, then its information content can be (partially or fully) reconstructed by its closest neighbors, and if there is a need, the reconstructed site may unload the traffic facing them.>