An expert system based load monitoring and scheduling system for load balancing in distributed systems

Abstract

It is argued that an expert system provides an ideal architecture for distributed control where each controller has to use information about the state of other decision makers to reach a common goal. It is suggested that an expert system is a viable tool for distributed problem solving where each solver has to monitor the state of other solvers and make decisions based on its approximate knowledge about other solvers. The authors illustrate this approach by taking the example of distributed load balancing. They present an expert-system-based architecture for load balancing in distributed systems. The system showed promising results during preliminary analysis. The system currently contains about 20 specialists and promises to be exceptionally robust and easily scalable. It is extremely fault tolerant and can function effectively in a changing environment.<>